A molecularly imprinted receptor for separation of testosterone and epitestosterone, based on a steroidal cross-linker

Article abstract of DOI:10.1016/j.steroids.2011.01.004

A series of molecularly imprinted polymers have been prepared and investigated as stationary phases in high performance liquid chromatography for the separation of testosterone and epitestosterone using non-polar mobile phases. The polymers were imprinted using 5α-dihydrotestosterone as template, and all retain testosterone more strongly than its 17α-OH epimer. The best polymer was prepared using trifluoromethylacrylic acid as functional monomer (interacting with the template via hydrogen bonds), divinylbenzene as 'inert' cross-linker, and chloroform as porogen. It also included a steroid-based cross-linker, which may interact with the template via van der Waals interactions to lend additional 'shape selectivity'. A 250 × 4.6 mm column packed with this polymer gave baseline resolution of testosterone and epitestosterone (15 μg each) in under 20 min. Preparation of the steroid based cross-linker included the selective reduction of 5α- dihydrotestosterone (17β-hydroxy-5α-androstan-3-one) to the 3α,17β-diol using K-selectride.

Full text of DOI:10.1016/j.steroids.2011.01.004

Steroids 76 (2011) 478–483
Contents lists available at ScienceDirect
Steroids
journal homepage: www.elsevier.com/locate/steroids
A molecularly imprinted receptor for separation of testosterone
and epitestosterone, based on a steroidal cross-linker
Ivana Gavrilovic´ a, Karen Mitchellb, Alan D. Brailsforda, David A. Cowana,
Andrew T. Kicmana, Richard J. Ansellb,∗
a Drug Control Centre, King’s College London, London SE1 9NH, United Kingdom
b School of Chemistry, University of Leeds, Leeds LS2 9JT, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of molecularly imprinted polymers have been prepared and investigated as stationary phases in
high performance liquid chromatography for the separation of testosterone and epitestosterone using
non-polar mobile phases. The polymers were imprinted using 5␣-dihydrotestosterone as template, and
all retain testosterone more strongly than its 17␣-OH epimer. The best polymer was prepared using
trifluoromethylacrylic acid as functional monomer (interacting with the template via hydrogen bonds),
divinylbenzene as ‘inert’ cross-linker, and chloroform as porogen. It also included a steroid-based cross-
linker, which may interact with the template via van der Waals interactions to lend additional ‘shape
selectivity’. A 250 × 4.6 mm column packed with this polymer gave baseline resolution of testosterone and
epitestosterone (15 ␮g each) in under 20 min. Preparation of the steroid based cross-linker included the
selective reduction of 5␣-dihydrotestosterone (17␤-hydroxy-5␣-androstan-3-one) to the 3␣,17␤-diol
using K-selectride.
Received 22 October 2010
Received in revised form 7 January 2011
Accepted 10 January 2011
Available online 19 January 2011
Keywords:
Molecular imprinting
Molecular recognition
Testosterone
Epitestosterone
5␣-Dihydrotestosterone
HPLC
© 2011 Elsevier Inc. All rights reserved.
1. Introduction
in functional group transformations, as selective adsorbents
in screening approaches, or solid-phase adsorbents for foren-
sic analysis. Molecularly imprinted polymers (MIPs) have also
been reported for closely related compounds including estradiol,
estrone and ethynylestradiol [24–34], for the analysis of these
endocrine-disrupting compounds in environmental samples, and
for cholesterol [7,35–42], for selective adsorption of this compound
and its detection in bodily fluids.
Molecular imprinting is a versatile approach for the prepara-
tion of polymeric materials capable of selective recognition of given
target molecules. The approach is based on complex formation
between a template molecule and polymerisable monomers. The
monomer–template interactions may be covalent or non-covalent
in nature. After polymerisation, the template is removed, leaving
vacant recognition sites, complementary in shape and functional
groups to the original template [1].
A number of groups have reported different strategies for
the molecular imprinting of testosterone [2–17] and related
steroids including methyltestosterone [18] androst-5-ene-3␤,17␤-
diol [19], castasterone [20], nandrolone [21], cortisol [22],
11␣-hydroxyprogesterone and corticosterone [23] and have sug-
gested applications for these polymers as protecting groups
For a new forensic assay for testosterone it was desirable to
develop an adsorbent capable of efficiently resolving testosterone
(T) and its 17␣-isomer, epitestosterone (E, Fig. 1). Although T
has been imprinted using non-conventional approaches such as
cyclodextrins [17] or porous silica [16], the majority of works on
T-imprinted polymers have employed conventional non-covalent
imprinting using T as template, methacrylic acid (MAA) as func-
tional monomer and ethyleneglycol dimethacrylate (EDMA) as
cross-linker. The only work which reports selectivity for T over E
is the recent one by Bui et al. [2], in which a methyltestosterone
(template)–MAA (functional monomer)–EDMA (crosslinker) poly-
mer showed a 20-fold selectivity for T over E in a competitive
radioligand binding assay. However, when used in a solid phase
extraction (SPE) protocol the polymer retained E almost as strongly
as T (reflecting the different conditions used in the two assays – the
radioligand binding assay probes only the strongest and most selec-
tive binding sites, while SPE uses a much larger number of sites on
the polymer, suggesting these weaker sites are much less selective).
Abbreviations: AnDMA, 5␣-androstane-3␣,17␤-dimethacryloxy ester; DHT, 5␣-
dihydrotestosterone,17␤-hydroxy-5␣-androstan-3-one; DVB, divinylbenzene; E,
epitestosterone; EDMA, ethylene glycol dimethacrylate; HPLC, high performance
liquid chromatography; MAA, methacrylic acid; MIP, molecularly imprinted poly-
mer; NIP, non-imprinted polymer; SPE, solid phase extraction; T, testosterone;
TFMAA, trifluoromethylacrylic acid.
∗
Corresponding author. Tel.: +44 113 3436415; fax: +44 113 3436565.
E-mail address: chmrja@leeds.ac.uk (R.J. Ansell).
0039-128X/$ – see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.steroids.2011.01.004

I. Gavrilovi´c et al. / Steroids 76 (2011) 478–483
479
Alfa Aesar (UK) and Fisher Scientific (UK). Chemicals were used as
received unless otherwise stated. All solvents were either labora-
tory reagent grade or HPLC grade. Dry solvents were obtained from
a solvent purification system (Pure Solv, Innovative Technologies,
USA). Basic aluminium oxide and silica gel for column chromatog-
raphy were purchased from Sigma Aldrich. Empty stainless steel
chromatography columns were purchased from Hichrom (UK).
2.2. Organic synthesis
5␣-Androstane-3␣,17␤-diol (2): DHT (2.00 g, 6.88 mmol) was
added to an oven dried two-neck round bottom flask. The flask
was fitted under argon and dry tetrahydrofuran (THF, 50 mL)
was added. When DHT was dissolved, the flask was cooled
with a dry ice/acetone bath and K-selectride (Potassium tri-sec-
butylborohydride, 1 M solution in THF, 10.32 mL, 10.32 mmol)
added dropwise. The reaction was kept at ∼−80 ◦C for 8 h and
then left running at room temperature overnight. Deionised water
(50 mL) was added to the flask and liquid–liquid extraction per-
formed with CHCl3 (3× 100 mL). The CHCl3 layers were collected
and purified with deionised water (3× 50 mL). The purified CHCl3
layer was dried with anhydrous MgSO4, filtered and concentrated
in vacuo. The white solid 2 was purified by recrystallisation from
CH2Cl2:CH3OH (95:5) (1H NMR consistent with the commercially
available compound, see Supplementary information)
5␣-Androstane-3␣,17␤-dimethacryloxy ester (AnDMA):
2
(2.92 g, 10 mmol), triethylamine (4.58 mL, 50 mmol), and 4-
dimethylaminopyridine (0.1752 g, 6 mmol) were stirred in dry
THF (150 mL) under argon and cooled to 0 ◦C using an ice bath.
Methacryloyl chloride (4.3 mL, 44 mmol) was added dropwise
and stirring continued at 0 ◦C for a further 2 h, then at room
temperature for another 24 h. The mixture was added to saturated
NaHCO3 solution (150 mL) and liquid–liquid extraction performed
with DCM (3× 150 mL). The DCM layers were collected, combined
and washed with saturated NaHCO3 solution (2× 150 mL) and
saturated NaCl solution (1× 150 mL). The DCM fraction was then
dried with MgSO4, filtered and 4-methoxyphenol (40 mg) was
added. Solvent was removed in vacuo to produce viscous yellow
oil (2.00 g). Hexane (300 mL) was added to the oil with shaking
and sonication, decanted into another flask and removed in vacuo
to produce a viscous yellow oil, which crystallised on drying
and storing at 5 ◦C to give a very pale yellow/white solid (2.45 g,
83.9%) m.p. 66–69 ◦C; found C 74.30%, H 9.20% (C27H40O4 requires
C 75.66%, H 9.41%); 1H NMR ı (500 MHz; CDCl3) 6.03 (2H, d,
2× CH2), 5.47 (2H, d, 2× CH2), 5.02 (1H, s, –CH at C3), 4.57
(1H, t, J 8.3 Hz, –CH at C17) 1.88 (3H, 2 s, 2× methacryl CH3),
0.76 (3H, 2s, –CH3 at C10 and C13) 2.25–0.67 (22H, m, –CH and
–CH2, remaining protons of steroid backbone); m/z (ESI) 446.3252
Fig. 1. Structures of (a) testosterone (T), (b) epitestosterone (E), (c) DHT, (d) AnDMA,
(e) the AnDMA analogue with 3␤-O-methacrylate. Structures calculated using PC
Spartan Pro, by semi-empirical minimisation using an AM1 force field.
17␤/17␣ selectivity has been observed between ␤-estradiol and
␣-estradiol on MIPs made from MAA/EDMA, both in competitive
radioligand binding assays [25,34] and when the polymers were
used as stationary phases in high performance liquid chromatog-
raphy (HPLC) [32], although improved separation in HPLC was
obtained with a MIP containing trifluoromethacrylic acid (TFMAA)
in place of MAA [33].
In this work, a series of MIPs have been prepared and eval-
uated for the separation of T and E in HPLC using a non-polar
organic mobile phase. The mobile phase was chosen to mimic the
wash step in the envisaged application of the MIPs in SPE: this
wash should elute E while T remains bound to the polymer. DHT
(5␣-dihydrotestosterone, 17␤-hydroxy-5␣-androstan-3-one) was
chosen instead of T as the template to avoid false results in SPE due
to template “bleeding”. Template bleeding occurs due to the small
amount of template remaining strongly bound to the polymer even
after applying extensive procedures for template removal. This
small amount of template bleeds slowly during the polymer appli-
cation and may interfere with analytical results [43]. To solve this
problem, a close structural analogue to the compound of interest is
chosen as the template. Bleeding of DHT from the polymer would
not interfere with the subsequent quantitative measurement of
T. In addition to evaluating different functional monomers and
‘inert’ cross-linkers, a functional cross-linker based on DHT (5␣-
androstane-3␣,17␤-dimethacryloxy ester, AnDMA, Fig. 1) has been
investigated in the expectation that van der Waals stacking inter-
actions might lend additional selectivity to the imprinted sites as
has been demonstrated by others in the case of cholesterol [40,44].
(M+NH4 requires 446.3265); ˛D = +15.9◦ (CHCl3)
+
2.3. Preparation of polymers
Divinylbenzene (DVB) was purified by passing through a plug
of basic alumina (Al2O3), then distilled under reduced pressure to
remove polymerisation inhibitors. MAA and EDMA were also dis-
tilled under reduced pressure. Polymers were prepared by bulk free
radical polymerisation. Amounts of chemicals used in the prepara-
tion of polymers are presented in Table 1.
Template (DHT), functional monomer (MAA or TFMAA) and
porogen (CHCl3 dried over molecular sieves) were weighed out
into a glass vial and the mixture sonicated until all material was
dissolved. Cross-linker (EDMA, DVB, AnDMA) was added and son-
ication was repeated until all material was dissolved. The initiator
2,2-dimethoxy-2-phenylacetophenone (25 mg) was added, the
vial was sonicated again and purged with argon for 3 min. The
vial was capped quickly and sealed with parafilm. Polymerisation
2. Experimental
2.1. General
Chemicals used in organic synthesis were obtained from Ster-
aloids (USA), Sigma Aldrich (UK), Acros Organics (UK), Fluka (UK),

480
I. Gavrilovi´c et al. / Steroids 76 (2011) 478–483
occurred overnight at 5 ◦C under UV light at 365 nm. MIP1 and
MIP2 were uncapped after the polymerisation and ground. Poly-
mers containing DVB were additionally heated for 72 h at 70 ◦C to
ensure the whole amount of DVB had polymerised and CHCl3 had
evaporated. Non-imprinted polymers (NIPs) as control polymers
were prepared the same way as imprinted polymers, but without
adding the template.
The polymer was crushed, ground in a mortar with pestle, and
fractionated by repeated sedimentation from acetone yielding par-
ticles of 5 < diameter < 30 ␮m. It might be anticipated that this
method would lead to a very wide distribution of particle sizes,
but in fact the size distribution is quite narrow (Fig. 2). MIP4a
was additionally sieved through a 20 ␮m sieve. Polymer particles
were packed into chromatography columns (50 × 4.6 mm, except
MIP4a) using an air-driven column packer (400 bar). The tem-
plate was removed from the polymer by flushing the column with
CH3OH:AcOH (8:2, v/v) overnight at 0.5 mL min−1. HPLC analysis
was performed on a Hewlett Packard 1090 liquid chromatograph
with Chemstation operating software. Injections of testosterone
and epitestosterone (3 mg mL−1 each) were performed man-
ually (5 ␮L injection loop) and chromatograms developed at
0.5 mL min−1 flow rate. Separation factors (˛) are reported as
(tRT − t0)/(tRE − t0) where tRT is the retention time for T, tRE is the
retention time for E and t0 is the void time. Imprinting factors
(IF) are reported as (tRT − t0) on the imprinted column divided by
(tRT − t0) on the non-imprinted control.
3. Results and discussion
3.1. MIPs with ‘inert’ cross-linker
Initially, DHT-imprinted polymers were produced using
a
conventional non-covalent imprinting approach with MAA or
TFMAA as functional monomer and only EDMA or divinylben-
zene (DVB) as cross-linker (Table 1, MIPs 1–3). The polymers were
made by UV-initiated polymerisation at 5 ◦C using chloroform as
porogen, crushed and 5–30 ␮m size particles packed into HPLC
columns (50 × 4.6 mm). Columns were connected to a HPLC sys-
tem with dichloromethane/AcOH (99:1, v/v) as mobile phase at
0.5 mL min−1. Testosterone and epitestosterone (15 ␮g each) were
injected and the measured retention times, separation factors and
imprinting factors are shown in Table 2.
The non-specific binding as measured by retention of steroids
on the NIPs increases in the order NIP1 (MAA/EDMA) < NIP3
(TFMAA/DVB) < NIP2 (TFMAA/EDMA). The TFMAA-based polymers
interact more strongly with the steroids because TFMAA is a bet-
ter H-bond donor than MAA: the stronger interaction of TFMAA (as
compared to MAA) with the carbonyl oxygen of the steroid was
confirmed by NMR titration (data not shown). NIP2 (EDMA cross-
linker) interacted more strongly with the steroids than NIP3 (DVB
cross-linker), because (despite the labelling of this cross-linker as
Fig. 2. Optical micrographs of (a) MIP3, (b) MIP4, and (c) MIP4a. Same magnification
in each case, scale bar is 80 ␮m.
Table 1
Amounts of template, monomer, cross-linker and solvent used in preparation of polymers. The same mass of DVB was used for MIP3 and 4 as EDMA in MIP1 and 2 (1.982 g),
the amount in mmol is calculated on the basis that DVB is 80% pure.
Polymer
Template DHT
(mmol)
Monomer (mmol)
Cross-linker (mmol)
Solvent
CHCl3 (mL)
MIP1
NIP1
MIP2
NIP2
MIP3
NIP3
MIP4
NIP4
MIP4a
0.625
–
0.625
–
0.625
–
0.625
–
MAA 2.5
MAA 2.5
EDMA 10
EDMA 10
EDMA 10
EDMA 10
DVB 12.1
DVB 12.1
DVB 12.1, AnDMA 1.25
DVB 12.1, AnDMA 1.25
DVB 77.4, AnDMA 8.00
2.80
2.80
2.67
2.67
3.05
3.05
3.05
3.05
19.52
TFMAA 2.5
TFMAA 2.5
TFMAA 2.5
TFMAA 2.5
TFMAA 2.5
TFMAA 2.5
TFMAA 16.0
4.00

I. Gavrilovi´c et al. / Steroids 76 (2011) 478–483
481
Table 2
Data obtained for separate injections of T or E (15 ␮g each) on MIPs and correspond-
ing NIPs. Imprinting factors (IFT) and separation factors (˛) calculated as described
in experimental section. Flow rate 0.5 mL min−1 except * 1.0 mL min−1
.
tRT (min)
tRE (min)
˛
IFT
DCM/AcOH, 99:1
MIP1
NIP1
MIP2
NIP2
MIP3
NIP3
MIP4
NIP4
1.62
1.31
5.23
3.00
3.78
1.63
2.41
1.18
1.44
1.29
4.12
3.05
2.31
1.61
1.92
1.17
1.53
–
1.37
–
2.21
–
1.60
–
2.48
–
2.17
–
5.06
–
16.38
–
DCM/AcOH, 99.9:0.1
MIP3
NIP3
MIP4
NIP4
12.64
2.69
7.86
1.64
5.84
2.66
3.94
1.58
2.43
–
2.38
–
7.26
–
12.52
–
DCM/AcOH, 99.5:0.5
MIP4a*
12.49
5.41
3.62
–
‘inert’) the steroid hydroxyl can also form weak hydrogen bonds
with the EDMA ester group.
None of the NIPs gave a significant difference in the retention of
T and E, but all of the MIPs did so. T bears more structural similarity
to the template, DHT, than E does, since T and DHT share the same
configuration at the 17-OH. Hence, T binds more strongly than E
does to the imprinted sites. MIP3 (TFMAA/DVB) gave the best sep-
aration of T and E (Fig. 3a). DHT interacts more with TFMAA in the
pre-polymerisation mixture than it does with MAA, creating more
and better recognition sites in the resulting MIP than when MAA is
used. As noted by Yilmaz et al., DVB as cross-linker may work bet-
ter than EDMA in combination with TFMAA, since TFMAA may also
interact with EDMA in the pre-polymerisation mixture such that
the number of TFMAA-template complexes is effectively reduced
[45]. Since TFMAA/DVB gave the best separation, further polymers
were based on this combination.
3.2. MIPs with ‘functional’ cross-linker
The separation of T and E achieved on MIP 3 was still not
adequate for the intended application. To lend further shape-
selectivity to the imprinted binding sites, the novel cross-linker
androsterone dimethacrylate (AnDMA) was designed and synthe-
sised (Fig. 1d and Scheme 1). It was envisaged that AnDMA would
interact with DHT in the pre-polymerisation mixture via van der
Waals stacking interactions. Such interactions are well-known in
cholesteric liquid crystals such as cholesteryl benzoate [46], in
aqueous assemblies of bile acids [47] and in organogels formed
by synthetic lithocholate [48] and cholesterol derivatives [49,50].
Organogelating androstane derivatives are less common but are
known [51,52].
Sellergren et al. [40] and Gore et al. [44] reported the
imprinting of cholesterol using functional monomers and cross-
linkers containing cholesterol covalently linked to polymerisable
methacrylate or itaconate groups. Sellergren’s MIPs also contained
MAA: when rebinding of cholesterol was measured in MeOH/H2O
cholesterol bound more strongly and selectively to the MIPs con-
taining the cholesterol-based crosslinker than to the MIPs which
did not. Gore’s MIPs contained only cholesterol based monomers
and crosslinkers: when rebinding was assessed in THF/H2O some
selectivity was observed, but the polymers were not compared to
conventional ones. Both Sellergren and Gore et al. attributed the
success oftheir polymers toapolar interactions between the choles-
terol template and cholesteryl monomer/cross-linker during the
imprinting process.
Fig. 3. Chromatograms for separate injections of T or E (15 ␮g each) (a) on MIP3
or NIP3 in DCM/AcOH, 99:1, (b) on MIP3 or NIP3 in DCM/AcOH, 99.9:0.1, (c) on
MIP4 or NIP4 in DCM/AcOH, 99:1, (d) on MIP4 or NIP4 in DCM/AcOH, 99.9:0.1. Each
chromatogram is the average trace for two separate injections.
To stack most efficiently with DHT, the steroid bis(methacrylate)
cross-linker here should have the same 5␣,17␤ configuration as
DHT. Molecular modelling suggested further that the 3␣ isomer
was more complementary to DHT, and to T, than the 3␤ (Fig. 1d
vs. e). 5␣-Androstan-3␣,17␤-diol (Scheme 1, 2) is commercially
available, but too expensive to be used as a precursor for a cross-
linker required in multigram quantities. Reduction of androsterone
(3␣-hydroxy-5␣-androstan-17-one) with NaBH4 in dioxane/water
was reported to give the 3␣,17␤-diol (2) selectively [53] and in our
hands this reaction was successful also in THF. A cheaper route to
2, however would begin from DHT (Scheme 1, 1) instead. Selec-
tive biotransformations of 1 → 2 have been published [54–56], but
were also rejected because of the scale required. NaBH4 reduc-
tion or hydrogenation of DHT gave mixtures of isomers [56,57].
K-selectride (potassium tri-sec-butylborohydride) has been shown

482
I. Gavrilovi´c et al. / Steroids 76 (2011) 478–483
Fig. 4. Chromatogram for injection of a mixture of T, E, methyltestosterone and 4-
androstene-3,17-dione on MIP4a, eluted at 1.0 mL min−1 using DCM/AcOH, 99.5:0.5
(v/v) as mobile phase.
ety of steroids using DCM/AcOH, 99.5:0.5 (v/v) as mobile phase,
at 1.0 mL min−1. The resulting chromatogram is shown in Fig. 4,
where the identities of the steroids were confirmed by separate
injections. The MIP displays an excellent separation of E and T,
although methyltestosterone is bound quite strongly and not com-
pletely separated from T. The identical configuration of the –OH
groups in T and methyltestosterone probably enables the latter
to fit into many of the recognition sites despite its extra methyl
group.
Scheme 1. Synthesis of AnDMA (3) from DHT (1) via 5␣-androstan-3␣,17␤-diol (2).
(a) K-selectride, THF, −80 ◦C. (b) Methacryloyl chloride, DMAP, TEA, DCM, 0 ◦C → RT.
to be a useful reagent for selective reduction of other 5␣-steroid-
3-ones to the 3␣-alcohols [58–60], although until now, use of this
reagent has not been reported for the reduction of DHT. In our hands
this reaction in THF at −80◦ C proceeded smoothly in high yield
giving exclusively the desired 3␣,17␤-diol 2 (see Supplementary
information for 1H NMR). 2 was then reacted with methacryloyl
chloride giving AnDMA (Scheme 1, 3) which could be purified from
hexane.
MIP4 was prepared containing AnDMA in addition to the com-
ponents of MIP3. Since the molecular weight of AnDMA is much
greater than that of DVB, a polymer in which AnDMA replaces
the DVB as cross-linker, would effectively be more lightly cross-
linked, and this was expected to be detrimental to the formation
of recognition sites. Hence, the amount of DVB used was not
changed. The amount of AnDMA used corresponds to a 2:1 ratio
of AnDMA:template and this was expected to be sufficient to indi-
cate whether AnDMA did improve the quality of recognition sites
or not.
MIP4 and NIP4 (containing AnDMA) did not retain T or E as
strongly as MIP3 and NIP3 (Table 2 and Fig. 3c and d). The lower
retention on the NIP indicates that some of the TFMAA may have
interacted with the AnDMA during polymerisation (similarly to
the interaction of TFMAA–EDMA), making it unavailable to interact
with the steroids when used in HPLC. Further, the separation factor
˛ for T and E on MIP4 was lower than that on MIP3. Nonetheless,
the imprinting factor IF on MIP4 was significantly better than on
MIP3, indicating the formation of more recognition sites in MIP4,
and the chromatograms showed that the peaks due to T and E were
better resolved on MIP4, due to the improved peak shape. The latter
effect may be due to a more homogeneous population of binding
sites. In a weaker mobile phase comprising DCM/AcOH, 99.9:0.1
(v/v), MIP4 was capable of almost baseline resolution of T and E,
while MIP3 gave a good separation but poor resolution, due to the
extremely broad peak for T. MIP3 seems to contain a small popula-
tion of strong binding sites, which are easily overloaded, while the
sites in MIP4 are more abundant and more homogeneous.
A larger batch of MIP4 (MIP4a) was produced and particles
sieved and then sedimented to yield a more homogeneous pop-
ulation of particles (Fig. 2) which were then packed into a longer
HPLC column (250 × 4.6 mm) and used for the separation of a vari-
4. Conclusion
The application of MIPs for steroid testing is relatively novel,
though its potential has recently been shown, most particularly in
the work of Bui et al. [2]. For some applications, a polymer such
as that of Bui et al. which binds T and E nearly equally may be
sufficient, while for others, separation of T and E is desirable. The
optimised polymer developed here shows baseline separation of
T and E in HPLC analysis, and in future work will be investigated
in other applications including molecularly imprinted solid-phase
extraction (MISPE).
Acknowledgments
The authors are very grateful to the World Anti-Doping Agency
for providing the funds for the project, which this work is part
of (Confirmation of doping with natural androgens by isotope
ratio mass spectrometry; simplifying the analytical procedure and
increasing the evidential power). Ivana Gavrilovic´ is also indebted
to King’s College London for meeting a substantial part of the costs
of her PhD registration fees.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.steroids.2011.01.004.
References
[1] Alexander C, Andersson HS, Andersson LI, Ansell RJ, Kirsch N, Nicholls IA, et al.
Molecular imprinting science and technology: a survey of the literature for the
years up to and including 2003. J Mol Recognit 2006;19:106–80.
[2] Bui BTS, Merlier F, Haupt K. Toward the use of a molecularly imprinted polymer
in doping analysis: selective preconcentration and analysis of testosterone and
epitestosterone in human urine. Anal Chem 2010;82:4420–7.
[3] Gadzala-Kopciuch R, Ricanyova J, Buszewski B. Isolation and detection of
steroids from human urine by molecularly imprinted solid-phase extraction
and liquid chromatography. J Chromatogr B 2009;877:1177–84.
[4] He CY, Long YY, Pan JL, Li K, Liu F. Molecularly imprinted silica prepared with
immiscible ionic liquid as solvent and porogen for selective recognition of
testosterone. Talanta 2008;74:1126–31.

I. Gavrilovi´c et al. / Steroids 76 (2011) 478–483
483
[5] He CY, Long YY, Pan JL, Li K, Liu F. A method for coating colloidal particles with
molecularly imprinted silica films. J Mater Chem 2008;18:2849–54.
[34] Ye L, Weiss RG, Mosbach K. Synthesis and characterisation of molecularly
imprinted microspheres. Macromolecules 2000;33:8239–45.
[35] Villar P, Whitcombe MJ, Vulfson EN. Matrix effects on the selectivity of a
cholesterol-imprinted polymer. Polymer 2007;48:1483–9.
[36] Ciardelli G, Borrelli C, Silvestri D, Cristallini C, Barbani N, Giusti P. Supported
imprinted nanospheres for the selective recognition of cholesterol. Biosen Bio-
electron 2006;21:2329–38.
[37] Boonpangrak S, Whitcombe MJ, Prachayasittikul V, Mosbach K, Ye L. Prepara-
tion of molecularly imprinted polymers using nitroxide-mediated living radical
polymerization. Biosen Bioelectron 2006;22:349–54.
[38] Spizzirri UG, Peppas NA. Structural analysis and diffusional behavior of molec-
ularly imprinted polymer networks for cholesterol recognition. Chem Mater
2005;17:6719–27.
[6] Salvador JP, Estevez MC, Marco MP, Sanchez-Baeza F.
A new methodol-
ogy for the rational design of molecularly imprinted polymers. Anal Lett
2007;40:1294–306.
[7] Sreenivasan K. Molecularly imprinted polyacrylic acid containing multiple
recognition sites for steroids. J Appl Polym Sci 2001;82:889–93.
[8] Sreenivasan K. On the application of molecularly imprinted poly(HEMA) as a
template responsive release system. J Appl Polym Sci 1999;71:1819–21.
[9] Sreenivasan K. Effect of the type of monomers of molecularly imprinted poly-
mers on the interaction with steroids. J Appl Polym Sci 1998;68:1863–6.
[10] Rachkov AE, Cheong SH, El’skaya AV, Yano K, Karube I. Molecularly imprinted
polymers as artificial steroid receptors. Polym Adv Technol 1998;9:511–9.
[11] Cheong SH, Rachkov AE, Park JK, Yano K, Karube I. Synthesis and binding prop-
erties of a noncovalent molecularly imprinted testosterone-specific polymer. J
Polym Sci Polym Chem 1998;36:1725–32.
[39] Kugimiya A, Kuwada Y, Takeuchi T. Preparation of sterol-imprinted poly-
mers with the use of 2-(methacryloyloxy)ethyl phosphate. J Chromatogr A
2001;938:131–5.
[12] Sreenivasan K. Application of molecularly imprinted polymer as a drug retain-
ing matrix. Angew Makromol Chem 1997;246:65–9.
[13] Sreenivasan K. A note on the selectivity of gamma-radiation polymerised
molecularly imprinted poly (HEMA). Polym Gels Netw 1997;5:17–22.
[14] McNiven S, Yokobayashi Y, Cheong SH, Karube I. Enhancing the selectivity of
molecularly imprinted polymers. Chem Lett 1997:1297–8.
[15] Cheong SH, McNiven S, Rachkov K, Levi R, Yano K, Karube I. Testosterone recep-
tor binding mimic constructed using molecular imprinting. Macromolecules
1997;30:1317–22.
[16] He CY, Liu F, Li KA, Liu HW. Molecularly imprinted polymer film grafted
from porous silica for selective recognition of testosterone. Anal Lett
2006;39:275–86.
[17] Sreenivasan K. Synthesis and evaluation of a beta cyclodextrin-based molecu-
larly imprinted copolymer. J Appl Polym Sci 1998;70:15–8.
[18] Yang ML, Gu WC, Sun L, Zhang F, Ling Y, Chu XG, et al. Study on the molec-
ularly imprinted polymers with methyl-testosterone as the template. Talanta
2010;81:156–61.
[19] Alexander C, Smith CR, Whitcombe MJ, Vulfson EN. Imprinted polymers as pro-
tecting groups for regioselective modification of polyfunctional substrates. J Am
Chem Soc 1999;121:6640–51.
[20] Kugimiya A, Matsui J, Abe H, Aburatani M, Takeuchi T. Synthesis of castas-
terone selective polymers prepared by molecular imprinting. Anal Chim Acta
1998;365:75–9.
[40] Sellergren B, Wieschemeyer J, Boos KS, Seidel D. Imprinted polymers for
selective adsorption of cholesterol from gastrointestinal fluids. Chem Mater
1998;10:4037–46.
[41] Whitcombe MJ, Rodriguez ME, Villar P, Vulfson EN. New method for the intro-
duction of recognition site functionality into polymers prepared by molecular
imprinting: synthesis and characterization of polymeric receptors for choles-
terol. J Am Chem Soc 1995;117:7105–11.
[42] Hishiya T, Shibata M, Kakazu M, Asanuma H, Komiyama M. Molecularly
imprinted cyclodextrins as selective receptors for steroids. Macromolecules
1999;32:2265–9.
[43] Ellwanger A, Berggren C, Bayoudh S, Crecenzi C, Karlsson L, Owens PK, et al.
Evaluation of methods aimed at complete removal of template from molecu-
larly imprinted polymers. Analyst 2001;126:784–92.
[44] Gore MA, Karmalkar RN, Kulkarni MG. Enhanced capacities and selectivities
for cholesterol in aqueous media by molecular imprinting: role of novel cross-
linkers. J Chromatogr B Anal Technol Biomed Life Sci 2004;804:211–21.
[45] Yilmaz E, Mosbach K, Haupt K. Influence of functional and cross-linking
monomers and the amount of template on the performance of molecularly
imprinted polymers in binding assays. Anal Commun 1999;36:167–70.
[46] Schenck R. Experiments on crystalline liquids. Z Phys Chem 1898;27:167.
[47] Terech P, de Geyer A, Struth B, Talmon Y. Self-assembled monodisperse steroid
nanotubes in water. Adv Mater 2002;14, 495+.
[48] Valkonen A, Lahtinen M, Virtanen E, Kaikkonen S, Kolehmainen E. Bile acid
amidoalcohols: simple organogelators. Biosen Bioelectron 2004;20:1233–41.
[49] Amaike M, Kobayashi H, Shinkai S. New organogelators bearing both sugar and
cholesterol units: an approach toward molecular design of universal gelators.
Bull Chem Soc Jpn 2000;73:2553–8.
[21] Percival CJ, Stanley S, Braithwaite A, Newton MI, McHale G. Molecular
imprinted polymer coated QCM for the detection of nandrolone. Analyst
2002;127:1024–6.
[22] Baggiani C, Giraudi G, Trotta F, Giovannoli C, Vanni A. Chromatographic
characterization of a molecular imprinted polymer binding cortisol. Talanta
2000;51:71–5.
[50] Zinic M, Vogtle F, Fages F. Cholesterol-based gelators. Top Curr Chem
2005;256:39–76.
[23] Ramstrom O, Ye L, Krook M, Mosbach K. Screening of a combinatorial steroid
library using molecularly imprinted polymers. Anal Commun 1998;35:9–11.
[24] Bravo JC, Fernandez P, Durand JS. Flow injection fluorimetric determi-
[51] Terech P, Friol S. Rheometry of an androstanol steroid derivative paramag-
netic organogel methodology for a comparison with a fatty acid organogel.
Tetrahedron 2007;63:7366–74.
nation of beta-estradiol using
2005;130:1404–9.
[25] Bui BTS, Belmont AS, Witters H, Haupt K. Molecular recognition of endocrine
disruptors by synthetic and natural 17 beta-estradiol receptors: a comparative
study. Anal Bioanal Chem 2008;390:2081–8.
[26] Dirion B, Schillinger E, Sellergren B. Development of a high throughput synthe-
sis technique for the optimization of MIPs for 17 beta-estradiol. Mater Res Soc
Symp Proc 2004;787:53–60.
[27] Le Noir M, Plieva F, Hey T, Guieysse B, Mattiasson B. Macroporous molecularly
imprinted polymer/cryogel composite systems for the removal of endocrine
disrupting trace contaminants. J Chromatogr A 2007;1154:158–64.
[28] Rachkov A, McNiven S, Cheong SH, El’Skaya A, Yano K, Karube I. Molec-
ularly imprinted polymers selective for beta-estradiol. Supramol Chem
1998;9:317–23.
a
molecularly imprinted polymer. Analyst
[52] Terech P, Furman I, Weiss RG, BouasLaurent H, Desvergne JP, Ramasseul R. Gels
from small molecules in organic solvents: structural features of a family of
steroid and anthryl-based organogelators. Faraday Dis 1995;101:345–58.
[53] McDermott IR, Robinson CH, DeKlerk DP. Studies on the effect of
5a-androstane-3a,17a-diol in the canine prostate in vivo. Steroids
1978;31:511–21.
[54] Ghraf R, Lax ER, Oza S, Schriefers H, Wildfeuer A, Haferkamp O. Transformation
of C18- C19- and C21-steroids by cultures of Candida albicans. J Steroid Biochem
1975;6:1531–6.
[55] Draczynska B, Tlomak E, Dmochowska-Gladysz J, Siewinski A. Microbiolog-
ical transformations. XVI. Transformation of 5Î - and 5Î2-dihydro and 1-
and 6-dehydro derivatives of testosterone and androstenedione by means
of Rhodotorula mucilaginosa strain. Bull Acad Pol Sci, Ser Sci Chim 1983;30:
13–21.
[29] Wang S, Xu ZX, Fang GZ, Zhang Y, He JX. Separation and determination of
estrone in environmental and drinking water using molecularly imprinted solid
phase extraction coupled with HPLC. J Sep Sci 2008;31:1181–8.
[30] Watabe Y, Kubo T, Nishikawa T, Fujita T, Kaya K, Hosoya K. Fully automated
liquid chromatography-mass spectrometry determination of 17 beta-estradiol
in river water. J Chromatogr A 2006;1120:252–9.
[31] Wei ST, Mizaikoff B. Binding site characteristics of 17 beta-estradiol imprinted
polymers. Biosen Bioelectron 2007;23:201–9.
[32] Ye L, Yu YH, Mosbach K. Towards the development of molecularly imprinted
artificial receptors for the screening of estrogenic chemicals. Analyst
2001;126:760–5.
[56] Prome D, Prome JC, Stahl D. Distinction between 17-epimeric hydroxy steroids
of the 3,17-dioxygenated androstane series by chemical ionization. Org Mass
Spectrom 1985;20:525–9.
[57] Parr MK, Zapp J, Becker M, Opfermann G, Bartz U, Schanzer W. Steroidal
isomers with uniform mass spectra of their per-TMS derivatives: synthesis
of 17-hydroxyandrostan-3-ones, androst-1- and -4-ene-3,17-diols. Steroids
2007;72:545–51.
[58] Gondos G, Orr JC. Stereoselective and regioselective reduction of steroid
ketones by potassium tri(R,S-S-butyl)borohydride. J Chem Soc, Chem Commun
1982:1239–40.
[59] Gondos G, McGirr LG, Jablonski CR, Snedden W, Orr JC. The reduction of steroid
2-alpha-fluoro-4-en-3-ones. J Org Chem 1988;53:3057–9.
[33] Zhu QJ, Tang J, Dai J, Gu XH, Chen SW. Synthesis and characteristics of imprinted
17-beta-estradiol microparticle and nanoparticle with TFMAA as functional
monomer. J Appl Polym Sci 2007;104:1551–8.
[60] Contreras R, Mendoza L. The reduction of 5a-cholestan-3-one and 5b-
cholestan-3-one by some boranes and hydroborates. Steroids 1979;34:121–4.