Characterizing Active Pharmaceutical Ingredient Binding to Human Serum Albumin by Spin-Labeling and EPR Spectroscopy

Article abstract of DOI:10.1002/chem.201601810

Drug binding to human serum albumin (HSA) has been characterized by a spin-labeling and continuous-wave (CW) EPR spectroscopic approach. Specifically, the contribution of functional groups (FGs) in a compound on its albumin-binding capabilities is quantit

Full text of DOI:10.1002/chem.201601810

DOI: 10.1002/chem.201601810
Full Paper
&
Structure–Activity Relationships
Characterizing Active Pharmaceutical Ingredient Binding to
Human Serum Albumin by Spin-Labeling and EPR Spectroscopy
Till Hauenschild,[a] Jçrg Reichenwallner,[a] Volker Enkelmann,[b] and Dariush Hinderberger*[a]
Abstract: Drug binding to human serum albumin (HSA) has
been characterized by a spin-labeling and continuous-wave
(CW) EPR spectroscopic approach. Specifically, the contribu-
tion of functional groups (FGs) in a compound on its albu-
min-binding capabilities is quantitatively described. Mole-
cules from different drug classes are labeled with EPR-active
nitroxide radicals (spin-labeled pharmaceuticals (SLPs)) and
in a screening approach CW-EPR spectroscopy is used to in-
vestigate HSA binding under physiological conditions and at
varying ratios of SLP to protein. Spectral simulations of the
CW-EPR spectra allow extraction of association constants (KA)
and the maximum number (n) of binding sites per protein.
By comparison of data from 23 SLPs, the mechanisms of
drug–protein association and the impact of chemical modifi-
cations at individual positions on drug uptake can be ration-
alized. Furthermore, new drug modifications with predicta-
ble protein binding tendency may be envisaged.
Introduction
In recent years, our group revealed, through a combined
CW-EPR and DEER spectroscopy approach, that the functional
(solution) structure of HSA, as seen from the distribution of
spin-labeled fatty acids (SL-FAs), features higher flexibility and
is more dynamic than that of its bovine analogue, bovine
serum albumin (BSA).[8,21,22,29] Extending our studies from
merely using SL-FAs, we present herein a new generalized ap-
proach to test ligand binding: we now utilize 2,2,6,6-tetrame-
thylpiperidine-1-oxyl (TEMPO) spin-labeled pharmaceuticals
(SLPs) as ligands and study SLP uptake by HSA through CW-
EPR spectroscopic analysis, similar to Cheng et al.[30]
Ligand uptake by biological and synthetic molecular transport-
ers, for example, human serum albumin (HSA),[1] has played
a central role in drug-delivery investigations for over 70 years.
Besides globulins and other sparingly occurring plasma pro-
teins, HSA is the major and one of the best studied transport
proteins in human blood plasma. It regulates many endoge-
nous processes and can serve as a drug carrier of many phar-
maceutical ligands;[2–7] this has made it a major research topic
in drug-delivery research.[2,8]
In addition to established methods for the investigation of
ligand binding to proteins, such as isothermal titration calorim-
etry (ITC);[9,10] Fçrster resonance energy transfer (FRET);[9–11]
XRD;[12] NMR spectroscopy;[13] and methods of EPR spectrosco-
py, such as continuous-wave (CW) EPR spectroscopy[14–18] or
nanoscale distance measurements by using double electron–
electron resonance (DEER) spectroscopy,[19–22] EPR-based meth-
ods that use paramagnetic molecules (spin probes) as ligands
offer the great advantage that in vitro ligand–protein binding
studies can be performed with high selectivity and under con-
trol of many physical and physiological parameters.[8,23–28]
After spin-labeling selected pharmacologically important,
active ingredients by use of Steglich esterification,[31] we can
discriminate between the amount of tightly HSA-bound and
freely tumbling SLPs. This is achieved by analyzing the rota-
tional slowdown of SLPs upon binding because the decrease
in rotational dynamics (as quantified by rotational correlation
times (tc)) can span up to about five orders of magnitude from
the 10 ps (fast isotropic tumbling) to the microsecond regime
(for tightly bound ligands).[32]
This study aims to understand how differences in chemical
substitution patterns of several chemically related pharmaceut-
icals influence the HSA-binding properties. By comparing two
or more chemically related pharmaceuticals differing in chemi-
cal substitutions at specific molecular positions, we aim to
draw precise conclusions as to which degree different func-
tional groups (FGs) contribute to protein binding. Furthermore,
we check the obtained binding parameters (herein: binding
sites per protein, KA) of the performed HSA/SLP binding studies
by comparing them with reference literature data of the un-
modified, EPR-silent pharmaceuticals (Figure 1).[3–7]
[a] T. Hauenschild, J. Reichenwallner, Prof. Dr. D. Hinderberger
Martin-Luther-Universitꢀt Halle-Wittenberg
Institute of Chemistry, Von-Danckelmann-Platz 4
06120 Halle (Saale) (Germany)
E-mail: dariush.hinderberger@chemie.uni-halle.de
[b] Dr. V. Enkelmann
Max Planck Institute for Polymer Research
Ackermannweg 10, 55128 Mainz (Germany)
Using CW-EPR spectroscopy, we first measure the ligand–
protein interactions under physiological conditions (pH 7.4,
Dulbecco’s phosphate-buffered saline (DPBS), 378C). Compared
Supporting information and ORCID from the author for this article are
available on the WWW under
http://dx.doi.org/10.1002/chem.201601810.
Chem. Eur. J. 2016, 22, 1 – 15
1
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Figure 1. Overview of known HSA–drug binding sites of several active pharmaceutical ingredients based on X-ray crystallographic studies.[3–5,7] Crystallograph-
ic analyses are usually carried out with the addition of fatty acids (FAs), that is, here as a HSA–myristate ligand complex (PDB ID: 2BXM). The binding sites for
drugs largely overlap with FA binding sites, as shown for the active pharmaceutical ingredients used in this study. Domains IIA and IIIA of HSA include two
primary drug-binding sites, known as Sudlow sites I and II.[33,34] Carbon atoms of myristic acid molecules are marked in light gray; the oxygen atoms are in
red.
with conventional ITC studies,[9,10] for CW-EPR spectroscopic
binding studies, nitroxide (NO) radicals, such as TEMPO, must
be chemically linked to the investigated drug. Ideally, in the
SLP, the chemical linkage of drug and radical should be dy-
namically flexible, so that the SLPs can be accommodated in
the binding sites with a minimum effect on the protein bind-
ing affinities (Figures S1 and S2 in the Supporting Information).
All other effects of impurities, such as denaturing influences on
HSA by traces of organic solvents, which is inevitable for dis-
solving the synthesized SLPs, can be eliminated by appropriate
reference measurements (Figures S3–S5 in the Supporting In-
formation). Generally, we can extract all binding parameters by
rigorously simulating[35,36] the recorded CW-EPR spectra. By
measuring series of samples with varying protein to ligand
(HSA/SLP) ratios, we can extract the spectral contributions, and
hence, fractions of bound and free SLPs as the thermodynami-
cally most insightful parameters. Using these numbers in an
analysis pioneered by Scatchard, one can extract the maximum
number (n) of binding sites per protein, the macroscopic bind-
ing association constant (KA), and the average binding affinities
(Baff) of the modified pharmaceuticals.[37–39]
Results and Discussion
Synthesis
We studied 23 active agents of 9 different pharmacologically
important drug classes (the medical use is given in parenthe-
ses). These comprise analgesics (pain relievers; Scheme 1), ace-
tylsalicylic acid (ASS; brand name: aspirin), salicylic acid (SAL),
ibuprofen (IBU), naproxen (NPX), paracetamol (PCM), diclofenac
(DCF), and indometacin (IMC); anticoagulants (bleeding coagu-
lation inhibitors; Scheme 2), phenprocoumon (PPC), warfarin
(WFR), and coumarin-3-carboxylic acid (CCS); the narcotic
(sleeping drug, anesthetic; Scheme 3) propofol (PPF); the uri-
costatic (uric acid inhibitors, against gout and hyperuricemia;
Scheme 4) febuxostat (FBS); the uricosurics (renal uric acid ex-
cretion accelerator; Scheme 5) probenecid (PBC) and benzbro-
marone (BBR); the cytostatic (cell growth and division inhibi-
tors; Scheme 6) chlorambucil (CBL); steroid hormones (specific
metabolic regulators; Scheme 7) cholesterol (CSO) and testos-
terone (TSO); and vitamins (intrinsic immune system regula-
tors; Scheme 8) d-a-tocopherol (DVE, vitamin E), d-(+)-biotin
(vitamin B7 (VB7)), niacin (vitamin B3 (VB3)), cholecalciferol (vi-
tamin D3 (VD3)), and ergocalciferol (vitamin D2 (VD2)); we also
include the (teratogenic) antiauxin 2,3,5-triiodobenzoic acid
(TIB), which is a plant growth regulation inhibitor (Scheme 9)
for its interesting chemical structure. These pharmacologically
active classes of molecules were examined with respect to
their binding behavior towards HSA. Spin labeling was carried
out by using either 4-hydroxy-TEMPO or 4-carboxy-TEMPO
(Scheme 10) through Steglich esterification.[31] Depending on
whether the hydroxyl or carboxyl groups of the selected drug
Our screening approach is then implemented by comparing
those binding parameters with values reported in the literature
and with the respective binding parameters of chemically relat-
ed pharmaceuticals in which the spin label was selectively in-
troduced at identical positions.
Herein, we present all related binding parameters and dis-
cuss each studied drug class individually to finally extract
some of the overarching design principles on an atomic level
that facilitate drug binding to HSA.
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Scheme 2. Molecular structures of the unmodified (RX =H) and SL anticoagu-
lants studied.
Scheme 3. Molecular structures of the unmodified (RX =H) and spin-labeled
narcotic studied.
Scheme 1. Molecular structures of the unmodified (RX =H) and spin-labeled
(SL) analgesics studied.
Scheme 4. Molecular structures of the unmodified (RX =H) and SL uricostat-
ics studied.
molecules were targeted for esterification/spin-labeling, the
corresponding TEMPO derivatives (Schemes 1–9) were em-
ployed.
The six-membered ring TEMPO-based NO radicals are very
stable at room temperature (RT) and at temperatures up to
ꢀ408C in dilute solutions.[40–42] Their characteristic stability was
also observed in this study because all SLPs could be easily re-
crystallized from temperatures up to 608C without loss of
signal, and a few SLPs could be stored at RT for extended peri-
ods of time. Nonetheless, to prevent potential chemical decay
during the labeling reaction, the Steglich method was used be-
cause it operates under very mild reactions conditions (syn-
thetic details can be found in the Supporting Information).[31]
To test for potential effects of DMSO, toluene, and DMF, for ex-
ample, on the solution structure of HSA, we performed refer-
ence measurements with up to 2% (v/v) of the solvents in buf-
fered[43] aqueous solution (see Figures S3–S5 and explanation
Scheme 5. Molecular structures of the unmodified (RX =H) and SL uricosurics
studied.
in the Supporting Information). In no case was any denaturing
effect of the solvent observed.
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
3
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Scheme 9. Molecular structures of all unmodified (RX =H) and SL antiauxins
studied.
Scheme 6. Molecular structures of the unmodified (RX =H) and SL cytostatic
studied.
Scheme 10. Schematic molecular and electronic structure and molecular
frame (tensor main axes) of TEMPO-based NO radicals. TEMPO (R=H), 4-car-
boxy-TEMPO (R=COOH), and 4-hydroxy-TEMPO (R=OH).
Scheme 7. Molecular structures of the two unmodified (RX =H) and SL ste-
roid hormones studied.
HSA binding influence of TEMPO derivatives used
To exclude that 4-hydroxy-TEMPO and 4-carboxy-TEMPO alone
bound to HSA themselves, corresponding reference measure-
ments were conducted. The measured CW-EPR spectra are
shown in Figures S1 and S2 in the Supporting Information and
show that the spin-labels alone do not interact with HSA.
Solvent influences on HSA in aqueous medium
Because some of the SLPs were not soluble in water, small
amounts of organic solvents had to be added to some sam-
ples.
Spectral composition of measured multicomponent spectra
It is possible to extract the relative concentrations of bound
(A), free (B), and micelle-bound (C) SLPs by multicomponent
spectral simulations[35,36,44] of the measured CW-EPR spectra.
Our analysis and approach is exemplified in Figure 2 for the
simulation of PPF-SL with HSA.
The complete simulations of the measured spectra (Fig-
ure S29 in the Supporting Information) were achieved by com-
bining the individual components (A, B, and C) as given by
Equation (1):
A þ B þ C ¼ x simðAÞ þ x simðBÞ þ x sim
ð1Þ
ðAÞ
ðBÞ
ðCÞ
ðCÞ
in which x(i) is the respective spectral contribution (percentage),
which is multiplied by sim(i), the simulated spectrum.
Figure 2 shows that these multiple-component spectra
(shown in red) excellently reproduce the experimental data (in
Scheme 8. Molecular structures of all unmodified (RX =H) and SL vitamins
studied.
black). The as-obtained fractions of the components are x(A)
0.53, x(B) =0.39, and x(C) =0.08.
=
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
4
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
ures S6 and S8 clearly show that the signal intensities of PPF-
SL and SAL-SL, in particular, are significantly increased in the
presence of HSA. This can be attributed to the micelle-break-
ing effect of HSA and subsequent ligand uptake of individual
SLPs by HSA, which yields much better CW-EPR spectra. Be-
cause ASS is the only well water-soluble molecule in our study,
its derivative ASS-SL also has high solubility in water (Fig-
ure S7) and does not form micelles. However, most of the SLPs
showed solubility characteristics similar to that of PPF-SL. Fur-
thermore, it seems that HSA has a critical limit in its micelle-
breaking effect, so that CW-EPR binding studies of some of the
SLPs could not be carried out, despite the determination of
a critical micelle concentration (CMC) in the range of 1 mm–
100 mm for each measured sample. The problem of micelle for-
mation could largely be circumvented by selecting incubation
times of 24 h (for all samples) and adjusting the total concen-
trations of the respective samples to either 100 or 600 mm,
which mark somewhat realistic lower and upper boundaries
for the blood concentrations of HSA, respectively.
CW-EPR binding studies of the SLPs
All binding capabilities/affinities to HSA were analyzed by
using the method pioneered by Scatchard.[37] All experimental
and simulated CW-EPR spectra and Scatchard analyses for all
SLPs are summarized Figures S9–S44 in the Supporting Infor-
mation. Herein, we restrict our detailed discussion to a selection
of systems (Figure 3, below, and subsequent discussion) and
summarize the binding parameters for all investigated SLPs in
Table 1.
According to standard conventions, the Scatchard plot is
constructed by using Equation (2):[38]
Figure 2. Spectral composition of individually simulated components A, B,
and C of PPF-SL with HSA at a ratio of 1:1 and a fixed protein concentration
of 100 mm. The most prominent spectral features of bound PPF-SL are high-
lighted in the insets.
½Bꢁ
½Fꢁ
1
1
ð2Þ
¼ ꢂ ½Bꢁ þ Bmax
KD KD
Mechanistic interpretation of SLP uptake by HSA
To test for the solubility of the synthesized SLPs, the freshly
prepared SLP samples were measured without HSA and with
HSA in a 1:1 ratio and at a constant total protein concentration
of 200 mm. In a comparison of CW-EPR spectra, in particular,
the normalized intensities and signal-to-noise ratios of the dif-
ferent SLP samples indicate similar solubilities in aqueous solu-
tion for all SLPs. The CW-EPR spectra of three selected SLPs
(ASS-SL, SAL-SL, and PPF-SL) and the corresponding reference
spectra without HSA are displayed in Figures S6–S8 in the Sup-
porting Information. Because all spectra are normalized and
plotted to scale, the clearly observable increase in signal inten-
sity when HSA is added because the amphiphilic SLPs (Fig-
ure S6, e.g., PPF-SL) form micelles in aqueous solution; these
do not show the typical three line spectrum of fast tumbling
NO radicals, but rather only a broad single line (exchange-nar-
rowed) spectrum that is due to high local concentrations in
the micelles and subsequent Heisenberg spin exchange. Fig-
in which [B] is the concentration of the bound component
(protein–SLP complex), [F] is the concentration of free SLP, KD
is the dissociation constant, and KA is the reciprocal value of
the dissociation constant (KDꢂ1). Usually, [B]/[F] is plotted
against [B] and one can obtain values of KD and Bmax from the
slope and x-axis intercept by least-squares fitting to a linear re-
lationship. Important binding information (maximum number
of binding sites per protein (Bmax ꢃn), association constants
(KA =KDꢂ1), and the average binding affinities (Baff)) are summar-
ized in Table 1 for each drug class. To determine Baff, which
could be extracted by suitable spectral simulations,[35–36,44] all
bound components were averaged over all measured SLP/HSA
CW-EPR spectra in the ratio of x:1. To prove the viability of our
CW-EPR spectroscopic approach, the binding sites per protein,
if reported in the literature,[3–7] of the corresponding unmodi-
fied pharmaceuticals are also given in Table 1.
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
effect and, on the other hand, can facilitate hydrogen bonding
to the protein, whereas the acetate group of ASS-SL has less
pronounced hydrogen-bonding activity.[45]
Table 1. Overview of all important binding parameters of all investigated
SLPs.[a]
[b]
Drug class
KA ꢄDKA [mꢂ1
]
Baff [%]
n
n*
nlit.
At a pH value of 7.4 and in combination with the relatively
high salt concentration of the DPBS buffer and measurement
temperature of 378C, a fraction of the acidic phenolate OH
groups may have been partly deprotonated (pKa(OH) ꢀ12.38),[46]
so that the free phenolate anion of SAL-SL may act as an even
more powerful hydrogen-bond acceptor. In addition, the ASS
acetate group has higher steric demands than that of the OH
group. IBU-SL and NPX-SL display similar binding behavior be-
cause both compounds are highly hydrophobic and approxi-
mately equal in molecular size. Among the SL analgesics, DCF-
SL und IMC-SL exhibit the highest binding affinities to the pro-
tein. This can be rationalized when taking into account that
electrostatic interactions between the polar CꢂCl bonds of
these two ligands and charged residues in the protein may sta-
bilize ligand-bound states. Furthermore, the high binding con-
stants of DCF-SL and IMC-SL suggest that the type of substitu-
tion pattern and the number of chloride atoms also appear to
play a role in the binding process.
analgesics
ASS-SL
SAL-SL
IBU-SL
NPX-SL
PCM-SL
DCF-SL
IMC-SL
6.33ꢄ0.31ꢁ103
1.48ꢄ0.07ꢁ104
2.90ꢄ0.03ꢁ104
2.17ꢄ0.17ꢁ104
5.36ꢄ0.29ꢁ102
6.24ꢄ0.24ꢁ104
4.85ꢄ0.28ꢁ104
48.6
66.3
73.9
72.2
35.9
79.0
78.3
1.98
2.01
2.01
1.97
2.01
1.99
2.03
2
2
2
2
2
2
2
2
2
2
2
–
–
2
anticoagulants
PPC-SL
WFR-SL
8.04ꢄ0.57ꢁ103
3.11ꢄ0.29ꢁ104
8.42ꢄ0.38ꢁ102
17.0
50.4
55.8
0.94
0.99
2.97
1
1
3
–
1
–
CCS-SL
narcotics
PPF-SL
9.46ꢄ0.69ꢁ103
55.8
–
1.98
–
2
–
2
–
uricostatics
FBS-SL
–
uricosurics
PBC-SL
BBR-SL
–
–
9.8
–
1.02
–
1
–
–
Our data indicate that electrostatic effects strongly influence
ligand-to-protein binding: PCM-SL has little if any electrostatic
component in binding to the protein and is thus bound
through hydrophobic interactions. Its association constant is
smaller, by two orders of magnitude, than the association con-
stants of DCF-SL and IMC-SL. A direct comparison of PCM-SL
with the chemically related analgesics ASS-SL and SAL-SL indi-
cates that the contribution of the acetamide group of PCM to
HSA binding affinity is significantly smaller than that of the
acetate group of ASS or even the OH group of SAL.
1.96ꢄ0.29ꢁ102
cytostatics
CBL-SL
5.05ꢄ0.24ꢁ104
69.5
2.02
2
–
steroids
TSO-SL
CSO-SL
4.39ꢄ0.37ꢁ104
72.0
–
1.15
–
1
–
1
–
–
vitamins
VB3-SL
VB7-SL
DVE-SL
VD3-SL
VD2-SL
4.75ꢄ0.28ꢁ102
33.6
20.0
–
–
–
2.02
1.03
–
–
–
2
1
–
–
–
–
–
–
–
–
4.69ꢄ0.42ꢁ102
This results in a binding affinity of PCM-SL that is smaller
than that of SAL-SL and ASS-SL by around 30 and 13%, respec-
tively. The weaker binding effect of PCM-SL to HSA can be ex-
plained by the formation of intramolecular hydrogen bonds
between the secondary amino group and neighboring carbon-
yl group of the acetamido group (Scheme 1).
–
–
–
antiauxins
TIB-SL
1.82ꢄ0.02ꢁ105
88.1
2.95
3
2
[a] KA: association constant; Baff: average binding affinity, averaged over
all bound components with respect to measured SLP/HSA CW-EPR spec-
tra in the ratio of x:1; n: maximum number of binding sites per protein
(n*ꢃrounded value). [b] All values were specified in accordance with ap-
propriate literature references.[3–7]
HSA binding of SL anticoagulants
For WFR-SL, we determined one binding site per HSA. This is
in agreement with the previously found value for the unmodi-
fied anticoagulant (WFR).[7] Of all investigated anticoagulants,
two different modified derivatives of the traditional coumarin
base, PPC-SL, WFR-SL (Scheme 2) and CCS-SL as the SL cou-
marin base, are chemically related and protein binding is
based primarily on hydrophobic effects and hydrogen bond-
ing. WFR-SL has the strongest affinity to HSA; PPC-SL shows
a significantly lower binding affinity (by 33.4%). The corre-
sponding association constant for PPC-SL is smaller than that
of WFR-SL by a factor of 10. The reason for this may be traced
back to the substitution pattern of both pharmaceuticals. WFR-
SL has a 3-oxo-1-phenylbutyl group, whereas PPC-SL contains
a 1-phenylpropyl group. The freely accessible 3-oxo-1-phenyl-
butyl group of WFR-SL is very likely to contribute to increased
flexibility of its coumarin backbone, so that the carbonyl (3-
oxo) group can act as a mild Lewis base and hydrogen-bond
HSA binding of SL analgesics
When inspecting Table 1, it becomes apparent that the
number of determined binding sites per protein for SL analge-
sics is in excellent agreement with the reported data for un-
modified analgesics.[3–5,7] A direct comparison of the two relat-
ed SL analgesics, ASS-SL and SAL-SL, shows that the free OH
groups of SAL-SL contribute a rather large increase in the aver-
age binding affinity (from 48.6% in ASS-SL to 66.3% in SAL-
SL). Therefore, the higher binding affinity of SAL-SL is mirrored
in its association constant, which is greater than that of ASS-SL
by more than a factor of two. This is probably because, on one
hand, SAL-SL interacts with HSA through the hydrophobic
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
acceptor. Hence, WFR-SL may bind to hydrogen-bond donors
in HSA, for example, amino acids, such as serine, or co-bound
water molecules. The 1-propyl group of the 1-phenylpropyl
group of PPC-SL has no hydrogen-bond acceptor function
(Scheme 2) and seems to be a poor hydrogen-bond donor
(binding affinities of WFR- and PPC-SL in Table 1). Indeed, simi-
lar to WFR-SL, PPC-SL shows only one binding site per protein,
so that the structural flexibility of both SLPs and their residual
mobility inside of the protein may not differ very much. For
reference purposes, CCS-SL, the coumarin basic body without
side-chain modification, was also measured. Compared with
PPC-SL and WFR-SL, the SL coumarin basic body displays two
interesting differences: CCS-SL has the weakest association
constant, but in our experiments occupies three binding sites
of HSA instead of one. These experimental findings may be ex-
plained by considering that the bare coumarin molecule, with-
out the additional side groups of PPC and WFR, is significantly
smaller than the two coumarin derivatives PPC and WFR
(Scheme 2). Compared with WFR-SL and PPC-SL, CCS-SL has
less structural flexibility: it is a carboxylic acid derivative of the
coumarin basic body and only the non-sterically hindered car-
bonyl group at position 2 may act as a hydrogen-bond accept-
or. This results in the lowest binding constant of all SL anticoa-
gulants, while simultaneously exploring the largest number
binding sites due to its smaller molecular size.
probably has a very low binding affinity to HSA because other-
wise upon HSA addition the EPR spectra of FBS-SL would show
components that stem from bound drug molecules (Figure S31
in the Supporting Information). We usually observe this in
other substance classes that form micelles, in particular, for
FAs.
HSA binding of SL uricosurics
PBC-SLP also displayed a large fraction of micelles of (85ꢄ5)%
and only (15ꢄ5)% freely tumbling PBC-SL (Figure S32 in the
Supporting Information).[47] In contrast, BBR-SL had a weak
binding affinity of 9.8%. One reason for the low protein bind-
ing may lie in its bulkiness, but, in particular, when taking into
account PBC-SL, too, the rather “kinked” and inflexible molecu-
lar structures of both uricosurics (Scheme 5), yielding large mo-
lecular volumes, may prevent accommodation in HSA to
a larger degree. The rather large number of heteroatoms and
FGs, in particular, both ortho-substituted aromatic bromine
atoms, may indicate another reason for the poor SLP uptake of
HSA, namely, increasing water solubility of BBR-SL in slightly
basic media (pH 7.4). To understand the effect of individual
FGs or side-chain modifications of the basic body of BBR on
HSA uptake and its solubility in water, further and more specif-
ic side-chain modifications may be necessary.
HSA binding of SL narcotics
HSA binding of SL cytostatics
We have experimentally determined two binding sites per HSA
for PPF-SL (Table 1), which coincides with the well-known liter-
ature value of the unmodified narcotic PPF.[7] PPF-SL is chemi-
cally related to the SL analgesics (ASS-SL, SAL-SL, PCM-SL) and
TIB-SL. All SLPs of these three substance classes (analgesics, an-
tiauxins, and narcotics) are aromatic phenyl esters with differ-
ent substitution patterns at the aromatic ring. PPF-SL has
a rather nonpolar character, similar to those of ASS-SL and TIB-
SL. The water solubility of PPF-SL is very low, which should
favor its uptake by HSA. However, the TEMPO ester of PPF pos-
sesses two bulky isopropyl groups, which are positioned at the
2,6-ortho positions relative to the ester bond of the aromatic
ring. Compared with ASS-SL, PPF-SL has a 7% higher binding
affinity to HSA. When inspecting the molecular structures of
ASS and PPF, it becomes apparent that, although having simi-
larities in bulkiness, the specific, symmetric ortho substitution
with isopropyl groups of PPF may lead to favorable binding in
the hydrophobic binding pockets of HSA. However, ASS is only
monosubstituted with an ortho-acetate group, which may lead
to binding to HSA through hydrogen bridges by which the
weaker hydrophobic interactions may be partly compensated
for.
CBL-SL has lower overall rigidity and more long-chain flexible
alkyl units with additional substitutions of chloride atoms at
the ends of the two alkyl chains. Therefore, we may study
a combination of potential interactions that contribute to the
protein binding of CBL-SL: On one hand, the negatively
charged/polarized chlorine atoms (and polar CꢂCl bonds) pre-
sumably interact electrostatically with the positive or negative
ends of the zwitterionic amino acid side chains. On the other
hand, the alkyl chains lead to a certain chain mobility and
structural flexibility, so that taken together the number of
binding sites per protein of CBL-SL is slightly increased. In ad-
dition to electrostatic effects, hydrophobic interactions play an
important role in the binding of CBL-SL to HSA. The binding of
CBL-SL to HSA could also finally be facilitated through the hy-
drogen-bond-accepting tertiary amino group. Overall, we mea-
sured a rather high binding affinity of around 70% and 2 bind-
ing sites per protein (Table 1).
As shown in Figure 3, the CW-EPR spectra of CBL-SL are
three-component spectra, although slight deviations between
simulations and recorded spectra are visible in the central
region. These deviations are due to a second micellar compo-
nent and can be removed completely when adding a micellar
fraction that has a distribution of spin exchange frequencies
(due to polydispersity in the micelles). Because the deviations
are in the range of less than 1% and in our eyes this fourth
species would not add additional insights, we have refrained
from using this fourth species and use the minimum of three
species. The micelle concentration increases from about 6% at
a ratio of 0.3:1 up to 44% at a ratio of 1:1. However, because
HSA binding of SL uricostatics
In this drug class, we could not detect HSA binding of FBS-SL.
At the concentrations needed for EPR spectroscopy, micelles
formed, even in the presence of HSA, as deduced from the
CW-EPR spectra.[47] This finding, in return, indicates that FBS-SL
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Figure 3. Measured (black) and simulated (red) CW-EPR spectra of CBL-SL at different CBL-SL/HSA ratios (x:1) and a fixed protein concentration of 100 mm.
we can distinguish all components quantitatively at any point,
this does not hamper the determination of the bound compo-
nents of the experimentally measured CW-EPR spectra (see
complementary explanations to the Scatchard-type analysis
[Eq. (2)] in the Supporting Information). For completeness, the
Scatchard plot of CBL-SL is shown in Figure 4. As previously
described in detail, we can derive the number of 2 binding
sites per protein at the x-axis intercept of the Scatchard plot
(200 mm) at a protein concentration of 100 mm.
forms micelles (Figure S35 in the Supporting Information)[47]
and does not show any binding to HSA in our concentration
regime. A direct comparison of both investigated steroid hor-
mones, TSO-SL and CSO-SL, shows that modifications in the 3-
and 17-positions of the steroid backbone significantly influ-
ence binding to the protein (Scheme 7). Therefore, CSO-SL is
not detectably bound by HSA, although one could speculate
that the hydrophobicity is increased by transfunctionalization
through a 1,5-dimethylhexyl group (at the 17-position of CSO,
compared with that of TSO), which should facilitate protein
binding. Modification with medium-length aliphatic chains at
the 17-position apparently drastically enhance micelle forma-
tion. In contrast, TSO merely has a hydroxyl group at this posi-
tion and in TSO-SL the amphiphilic spin label is attached at
this position. In CSO-SL, the spin label is chemically attached at
the 3-position, where TSO-SL has a keto group. The high bind-
ing constant of TSO-SL can be explained by, first, a smaller hy-
drophobic side-chain modification of the basic steroid struc-
ture (spin labeling), so that micelle formation is largely sup-
pressed and TSO-SL can be competitively bound by HSA (in
CSO-SL: highly hydrophobic dimethylhexyl group). Second,
TSO has a carbonyl group at the 3-position of the steroid back-
bone, which may act as a hydrogen-bond acceptor and form
hydrogen bridges with hydrogen-bond donors in the binding
sites of the protein (in CSO-SL, the TEMPO spin label is at-
tached as an carboxylic acid ester).
HSA binding of SL steroid hormones
From the binding data summarized in Table 1, it becomes clear
that the number of binding sites for TSO-SL is identical to the
value of TSO reported in the literature.[6] In contrast, CSO-SL
HSA binding of SL vitamins
All synthesized vitamins, VD3-SL, VD2-SL, and DVE-SL, show
strong micelle formation, so that the concentration-dependent
binding studies to HSA could not be carried out (see Fig-
ures S38–S40 and explanations in the Supporting Information).
Apparently, HSA could not exert its moderate micelle-breaking
Figure 4. Scatchard plot[32] of CBL-SL at different CBL-SL/HSA ratios and
a fixed protein concentration of 100 mm.
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
8
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
effect, which indicated that all three fat-soluble vitamins were
not bound by HSA in their SL form. A small binding of investi-
gated vitamins D and E to HSA cannot be excluded due to
a strong micelle formation, but the largest percentage of vita-
mins D and E are known to be bound to glyco- and lipopro-
teins.[48–52] As shown in Table 1, the two water-soluble vitamins,
VB3-SL and VB7-SL, do not display large association constants
either. Nevertheless, it cannot be excluded that both SL vita-
mins, if only slightly, are bound by HSA under physiological
conditions. Although most of the synthesized SLPs are com-
pounds with a rather amphiphilic or even nonpolar character,
compounds such as VB7-SL, which contain polar and nonpolar
groups, do not partition into the protein as strongly as more
hydrophobic compounds. The main reason for the poor pro-
tein binding of VB7-SL could be an intrinsic problem in molec-
ular design: the two NH groups of the 2-oxo-imidazolidine ring
can form intramolecular hydrogen bonds with the carbonyl (2-
oxo) group, so that the formation of intermolecular hydrogen
bonds may be severely hindered. Interestingly, this could indi-
cate the possibility of modifying small molecules that can
switch hydrogen-bonding networks from being intra- to inter-
molecular, and hence, facilitate binding to HSA, as already
found in responsive polymers.[28]
binding affinity to HSA.[8,13] Potentially, the actual maximum
number of binding sites attainable by TIB (also in the non-SL
form) may be three for binding without competition from
long-chain FAs. In accordance with our data on other halogen-
ated SLPs, such as BBR-SL, IMC-SL, DCF-SL, or CBL-SL, ionic/
electrostatic interactions play a crucial role for the protein
binding of TIB-SL. Relative to these aforementioned SLPs, TIB-
SL is even more strongly bound to HSA, which could be ex-
plained by the increased number and substitution pattern of
iodine atoms. Because the iodine atoms are more (negatively)
polarizable than chloride or bromine atoms and the amino
acid residues can be present as zwitterion in the protein, the
binding of TIB-SL is probably facilitated by increased electro-
static/dipole–dipole interactions. As shown in Figure 5, similar
to CBL-SL, the CW-EPR spectra of TIB-SL are pure three-compo-
nent spectra. The micelle concentration in the CW-EPR spectra
increases from 8% at a TIB-SL/HSA ratio of 1:1 to a value of
around 19% at a ratio of 1.3:1. From the x-axis intercept of the
Scatchard plot of TIB-SL (Figure 6), the number of TIB-SL bind-
ing sites per HSA molecule reaches a value of three. The rather
small molecular size/volume of TIB-SL may also contribute to
this rather high number of TIB-SL binding sites.
Conclusions
HSA binding of SL antiauxins
When discussing the large sets of data together, it can be
stated that all 23 synthesized SL compounds used for the CW-
EPR binding studies were obtained in sufficient quantities and
necessary purity. By comparison with corresponding data in
the literature, we can show that the HSA/SLP binding studies
conducted by EPR spectroscopy lead to the same number of
available binding sites as those in the non-SL cases (wherever
respective values were reported in literature), with the excep-
tion of TIB-SL, which has an even higher number of maximum
binding sites in the SL version (three vs. two reported in the
As observed from the results in Table 1, we determined three
binding sites per protein for TIB-SL. This is the only case in
which our experimentally determined value for the maximum
number of binding sites per protein (n) deviates from the liter-
ature value of two for the unmodified antiauxin (TIB).[7] Be-
cause the specific number of binding sites of unlabeled TIB
were determined in the presence of myristic acid, one may
speculate that, in previous studies,[7] some of the binding sites
of HSA were competitively blocked by FAs with rather strong
Figure 5. Measured (black) and simulated (red) CW-EPR spectra of TIB-SL at different TIB-SL/HSA ratios (x:1) and a fixed protein concentration of 100 mm.
Chem. Eur. J. 2016, 22, 1 – 15 www.chemeurj.org ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9
&
&
These are not the final page numbers! ÞÞ

Full Paper
easy-to-use CW-EPR spectroscopy, with its high sensitivity and
especially selectivity, we could further obtain precise data on
the protein binding affinities of the investigated pharmaceuti-
cals. Building on sparse EPR spectroscopic studies and
a wealth of other binding studies to HSA, our approach is to
combine a selection of pharmaceuticals that cover many struc-
tural and substitution motifs with spin-labeling and Scatchard-
type analysis. We could show that our method was very robust
against chemical impurities and traces of solvents (as opposed
to, e.g., calorimetric methods). It is remarkable to note that
spin-labeling with TEMPO seems to have only small effects on
the binding behavior to HSA. This is despite the fact that
TEMPO is a medium-sized heteroatom ring system that is even
comparable in size to some of the pharmaceutical molecules
(e.g., the analgesics ASS-SL, SAL-SL).
All in all, steric demand of the esterified TEMPO spin label
does not seem to be the critical issue because binding of the
pharmaceuticals in a native (i.e., similar to the unmodified mol-
ecule) fashion is not prevented, as our data show. This forms
an excellent basis for the development of our approach for
ligand binding studies.
Figure 6. Scatchard plot[32] of TIB-SL at different TIB-SL/HSA ratios and
a fixed protein concentration of 100 mm.
unlabeled case). One may speculate that modification with an
amphiphilic molecule, such as TEMPO, may enhance the bind-
ing affinity and number of binding sites of TIB-SL in an indirect
manner, but because our reference measurements with TEMPO
alone (see the Supporting Information) show no interaction of
the spin label alone with HSA, it is reasonable, although cir-
cumstantial, to argue that the actual maximum number of
binding sites of TIB may be three and not two, as reported in
a study with competitive binding from long-chain FAs. Using
A direct comparison between CW-EPR spectroscopy and ITC
(Table 2), which is a well-established method for determining
protein–ligand interactions,[53–58] indicates that our EPR-based
method is competitive.
Based on our experimental data (Table 1), one can deduce
that even small modifications of the backbone of chemically
related active ingredients in one drug class may lead to dis-
Table 2. Competitive comparison of the advantages and disadvantages of CW-EPR spectroscopy (this work) and ITC.[53–58]
CW-EPR
ITC
spin probe must be attached to investigated drug, but drug-linked spin
probe (TEMPO) shows no influence on protein binding (Figures S1 and S2)
no spin-labeling necessary
drug–protein interactions can be carried out under physiological conditions (pH 7.4, DPBS buffer, 378C)
background and/or baseline corrections are usually not required
background and/or baseline corrections are always necessary
suitable for aqueous and binary organic/aqueous systems!traces of differ- generally suitable for aqueous systems!traces of organic solvents lead to
ent organic solvents (ca. 2 vol%) show no denaturing effect on HSA (Figur- measurement inaccuracies (e.g., no control of protein denaturing effects,
es S3–S5)
possible destruction of sample cell, unexpected enthalpy changes)
rapid sample preparations, no equipment cleaning (thanks to capillary tube), time-consuming sample preparations (e.g., degassing of all samples), tedi-
short measuring times (few minutes), but time-consuming data analysis
ous equipment cleaning (injection unit/sample cell), long measuring times
(several hours) and data analysis
spectral simulations offer the possibility of extracting all relative concentra- separate determination of bound and micelle-bound states is not precisely
tions of bound, free, and micelle-bound states of all SLPs by rigorously sepa- possible, since both states are superimposed states!measured enthalpy
rating all spectral signatures; micelle-prone CW-EPR spectra can be detected changes always include a micellar component
in the final spectra (Figures 3 and 5)
sensitivity: detection of sample concentrations in the lower mm to upper nm sensitivity: detection of sample concentrations in the lower mm range
range (device and sample dependent)
required sample volume: a few microliter (ca. 50–100 mL) per complete mea- required sample volume: a few milliliter (ca. 1.5–2.5 mL) per measurement
surement series
series
information: association (KA) and dissociation constants (KD), maximum
information: association (KA) and dissociation constants (KD), maximum
number (n) of binding sites per protein, average binding affinity (Baff), spec- number (n) of binding sites per protein, Gibbs free binding energy (DG) of
tral composition of bound, free and micelle-prone states
binding process, the enthalpy (DH) and entropy (DS) of binding, reaction
stoichiometry (m)
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
10
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
tinctly different protein affinities. As a result, side-chain func-
tionalization with halogenated groups, such as iodine or chlo-
ride atoms, leads to an enhanced HSA–drug interactions. The
O- and especially N-acetylated groups, as in ASS-SL or PCM-SL,
lead to weaker binding to HSA. In contrast, the OH groups of
SAL-SL have a strong positive impact on HSA binding; this is
likely to be because OH groups can serve as hydrogen-bond
donors and acceptors or as a powerful hydrogen-bond accept-
or in the deprotonated state. Generally, it can be stated that al-
bumin shows a high affinity for small hydrophobic molecules,
which are modified with mostly negatively charged or (nega-
tively) polarizable groups. This is substantiated through strong
HSA binding of halogenated compounds, for example, TIB-,
DCF-, IMC-, and CBL-SL, or FAs. Therefore, it seems that hydro-
phobic and electrostatic effects are essential for significant pro-
tein binding. If the hydrophobic character of a molecule by
long-chain alkyl (nonpolar) side-chain modifications becomes
too large, the tendency for micelle formation due to poor
water solubility, to a certain degree, counteracts binding to
HSA.[2,59] For cases in which micellization is too strong (e.g.,
CSO-SL), no ligand–HSA interaction could be detected. As
a good example of the favorable combination of different ef-
fects (here: electrostatic and hydrophobic effects), CBL displays
better binding to HSA and only slight micelle aggregation,
whereas the micellar structures of DVE or CSO, for example,
are too stable and prevent interaction with HSA.
EPR spectroscopy as a powerful tool to characterize the impact
of individual chemical moieties and FGs on HSA binding. Un-
derstanding the effect of single changes in a molecular struc-
ture is part of ongoing projects in our group and may finally
aid the design of molecules or derivatives of known drugs
with a predictable binding to HSA after selective modifications,
even if the drugs showed no previous protein binding tenden-
cy. In the case of HSA, this is hoped to give enhanced reten-
tion and permeation (also called the EPR effect) of amphiphilic
or hydrophobic drugs in the blood stream by binding to HSA.
The approach is, of course, not restricted to this serum trans-
port protein, but, in principle, can be transferred to other
types of transporter, for example, globulins in the blood
serum.
Experimental Section
Sample preparation
Only the procedure for general sample preparation is described. A
detailed explanation of all sample preparations with respect to all
reference measurements and performed binding studies is system-
atically found in subsections ii–vii in the Supporting Information.
Procedures for general sample preparation: All 1.5 mL Eppendorf
Safe-Lock Tubesꢂ were autoclaved before use. All samples were pi-
petted together with an Eppendorf Researchꢂ pipette set and asso-
ciated standard pipette tips from the same manufacturer (Eppen-
dorf AG). The tubes described in subsections iii–vii in the Support-
ing Information were resealed after each pipetting step and the
sample solution was homogenized by corresponding vortexing
with a vortex mixer IKAꢂ Vortex Genius 3 instrument. As described
in subsections iv–vii in the Supporting Information, all prepared
samples included a proportion of organic solvent of 2 vol% and
were measured under physiological buffer conditions by means of
an X-band CW-EPR spectrometer. In addition, all sample solutions
(see subsections iv–vii in the Supporting Information) were each
prepared with 20 vol% of an 87% mixture of glycerol/water. This
served not only for cryoprotection of freshly made samples, but as-
sured identical viscosities, h. The readjustment of the pH value by
appropriate titration (see Tables S1–S5 in subsections iv–vii in the
Supporting Information) of all prepared samples was carried out
with the DPBS buffers[43] of various pH values described in subsec-
tion ii in the Supporting information. Unless stated otherwise in
the Supporting Information, all prepared samples were measured
at a temperature of 378C.
Based on the SLPs studied, we can identify some FGs (see
the Results and Discussion) that contribute to enhanced pro-
tein binding and are suitable for use in rational molecular
design. With a few exceptions (namely, ASS-SL/SAL-SL/TIB-SL,
WFR-SL/PPC-SL, or PCM-SL/PPF-SL; see Schemes 1–3 and 9 and
Table 1 for differences in binding affinities), it is beyond the
scope of this study to examine the exact impact of individual
FGs on otherwise identical compounds on HSA binding.
When inspecting the coincidence of the number of binding
sites per protein of the 23 SLPs (Table 1) and of their unmodi-
fied analogues (see Figure 1), it is justified to state that most
of the investigated SLPs, for example, ASS-SL, are bound in the
primary HSA domains IIA and IIIA, known as Sudlow sites I and
II (Figure 1), as previously known.[33,34] Binding of the SLPs or
unmodified pharmaceuticals in these binding sites may hence
be controlled through molecular design (e.g., by suitable
choice of one stereoselective center inside of a molecule back-
bone or side chain). Also, through the orientation of the side-
chain modifications/FGs inside of the protein (depending on
control of (S)- or (R)-enantiomer modifications), it seems to be
possible to influence or navigate the binding behavior of
a new synthetic pharmaceutical drug in either of these two, or
even more (see TIB-SL), binding sites of HSA. When inspecting
HSA binding of CBL-SL and TIB-SL (see Figures 3 and 5), it is
very interesting to see the development of the mono- to multi-
ple halogenation effect on protein binding. In the future, one
may envision designing specific pharmaceutical drugs with
a predictable, optimized protein binding tendency. Finally, our
results suggest that there is a subtle interplay of electrostatic,
hydrophobic, and hydrophilic effects for a strong HSA–drug as-
sociation. The present study can be seen as paving the way for
Technical details of the measured CW-EPR spectra
The modulation amplitude was set 0.1 mT. The magnetic field (B)
was centered at 334.6 mT at a microwave (MW) frequency of about
9.377 GHz, the sweep width was set at 11.90 mT, and the sweep
time was selected to be 60 s. The number of scans was selected to
be a reasonable compromise between measurement time and
signal-to-noise ratio (SNR). The number of scans was varied from 8
to 32, depending on the SNR. The gain (signal amplification) was
adjusted at a constant MW attenuation of 25 dB (MW power =
0.32 mW) for each measurement, so that the obtained CW-EPR
signal was clearly visible. All further CW-EPR technical details can
be found in the subsection on EPR spectroscopy in the Supporting
Information.
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
11
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
(FD, 8 kV): m/z (%): 654.0 (100) [M+]; elemental analysis calcd (%)
for C16H19I3NO3: C 29.38, H 2.93, N 2.14; found: C 29.51, H: 2.49, N:
2.19.
Characterization techniques
All characterization techniques, NMR spectroscopy, mass spectrom-
etry (MS), elemental analysis (EA), IR spectroscopy, X-ray structure
analysis, melting point determination, EPR spectroscopy, prepara-
tive column chromatography, and TLC, are described in detail in
the Supporting Information.
X-ray structure analysis: Four single crystals were obtained of the
twenty-three synthesized SLPs (SAL-SL, PCM-SL, PBC-SL and TIB-SL)
and these were examined by X-ray single-crystal structural analysis
(Figures 7–10). Consequently, all important crystallographic data of
the four investigated SLPs, such as bond lengths, bond angles,
crystal system, space group, and cell volume, are given in Ta-
bles S7–S14 in the Supporting Information.
Synthesis
All commercially available chemicals were purchased from Sigma–
Aldrich Chemie GmbH, ABCR GmbH & Co. KG, VWR International
GmbH, or ACROS Organics in their highest degree of purity and
were used without further purification. The corresponding suppli-
ers for all active pharmaceutical ingredients used are specified in
brackets in the following reaction descriptions. N,N’-Dicyclohexyl-
carbodiimide (DCC), 4-(dimethylamino)pyridine (DMAP), and
TEMPO derivatives 4-hydroxy-TEMPO and 4-carboxy-TEMPO were
purchased from Sigma–Aldrich Chemie GmbH (Taufkirchen, Germa-
ny). All solvents were of HPLC grade and purchased from Sigma–
Aldrich (DMSO, CHCl3), Acros Organics (DMF, CH2Cl2) or VWR Inter-
national GmbH (EtOH, n-hexane (n-Hex), ethyl acetate (EA), diethyl
ether (DEE), petroleum ether (PE), acetone (ACE), n-pentane,
MeCN) and used as received for TLC, column chromatography, and
recrystallization. The solvents required for chemical reactions were
dried by the respective standard methods and stored over the ap-
propriate molecular sieves when they were not available in pure
form from the corresponding suppliers. For PPC and FBS, prescrip-
tions for purely scientific purposes were obtained from PD Dr.
Volker Mailꢃnder at the MPI for Polymer Research in Mainz: Marcu-
marꢂ 3 mg tablets (Meda Pharma GmbH & Co. KG) and Adenuricꢂ
120 mg film-coated tablets (Berlin Chemie AG). All reactions were
carried out under argon acquired from Linde AG. In the following,
the synthesis of TIB-SL is described. All other reaction descriptions
of the investigated SLPs are described in the Supporting Informa-
tion.
Figure 7. Single-crystal structure of 2,2,6,6-tetramethylpiperidine-1-oxyl-4-yl
2-hydroxybenzoate (SAL-SL).
2,2,6,6-Tetramethylpiperidine-1-oxyl-4-yl 2,3,5-triiodobenzoate
(TIB-SL): Under inert gas conditions, TIB (199.9 mg, 0.40 mmol,
1 equiv; Sigma–Aldrich Chemie GmbH), 4-hydroxy-TEMPO
(103.3 mg, 0.60 mmol, 1.5 equiv), and DMAP (11.2 mg,
ꢅ0.09 mmol, 0.12 equiv) were dissolved in dry CH2Cl2 (15 mL) and
DMF (15 mL) at RT under constant stirring. The resulting red solu-
tion was cooled to 08C and DCC (107.3 mg, 0.52 mmol, 1.3 equiv)
in dry CH2Cl2 (2.5 mL) was slowly added dropwise. After 2 h at 08C
and further stirring for 24 h at RT, the reaction mixture could be
evaporated to dryness, after the appropriate reaction control (TLC:
Rf (n-Hex/EA (6:1))=0.36). Next, the crude product was dissolved in
CH2Cl2 (ꢅ10–15 mL), thereby the precipitated N,N’-dicyclohexylur-
ea (DCU) was filtered off and the obtained bright red filtrate con-
centrated to about 4 mL. After purification of the crude product by
column chromatography (SiO2 (0.063–0.2 mm); n-Hex/EA (6:1)), the
product was recrystallized from n-Hex to give TIB-SL, the TEMPO-
ester of TIB, as a red crystal-line solid (202.7 mg (0.31 mmol, 78%).
M.p. 155.58C; IR (KBr): v˜ =3099, 3076, 3001 (=C-H, m–w); 2987,
2981, 2966, 2956, 2935, 2925 (-CH2, w), 2902, 2864 (-C-H, m-w);
1722 (C=O, s), 1680, 1652 (C=C, w), 1554, 1540, 1518 (N-O, m),
1466, 1458, 1450, 1438 (-CH3- u. -CH2-def., m), 1392, 1377, 1367,
1361, 1352, 1311 (-CH3-def. and N-O, s–m), 1292, 1260, 1231 (C-O-
C, m–s), 1192, 1178, 1124, 1105, 1086, 1026 (C-O-C, s–m), 999, 980,
963, 941, 913, 898, 878, 867, 843 (=C-H-def. and -C-C, s–m); 799,
777, 735 (-CH2-def., m), 710, 686, 673, 600, 561, 553, 511 cmꢂ1 (-C-I,
m–w); ESR (X-Band, Mn-Std., DMSO, 258C): Axx =Ayy =17.3 MHz,
Azz =98.2 MHz; Aiso =44.3 MHz; giso =2.006269; tR =123.2 ps; MS
Figure 8. Single-crystal structure of N-acetyl-4-aminophenyl 2,2,6,6-tetrame-
thylpiperidine-1-oxyl-4-carboxylate (PCM-SL).
Figure 9. Single-crystal structure of 2,2,6,6-tetramethylpiperidine-1-oxyl-4-yl
4-(dipropylsulfamoyl)benzoate (PBC-SL).
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
12
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[12] A. A. Bhattacharya, T. Grꢄne, S. Curry, J. Mol. Biol. 2000, 303, 721–732.
[13] J. R. Simard, P. A. Zunszain, J. A. Hamilton, S. Curry, J. Mol. Biol. 2006,
361, 336–351.
[14] T. Hauenschild, Spinmarkierung pharmazeutischer Wirkstoffe sowie deren
Bindungseigenschaften an Serum Albumin vom Menschen, Diploma
Thesis, Johannes Gutenberg University Mainz, 2013; published by
MPG.PuRe: http://hdl.handle.net/11858/00-001M-0000-0015-1BD8-0.
[15] D. Tatlidil, M. Ucuncu, Y. Akdogan, Phys. Chem. Chem. Phys. 2015, 17,
22678–22685.
[16] M. D. Barratt, M. D. Rickwood, Biophys. Chem. 1984, 19, 69–73.
[17] M. Basset, E. M. Chambaz, G. Defaye, B. Metz, Biochimie 1978, 60, 715–
724.
[18] M. Basset, G. Defaye, E. M. Chambaz, Biochim. Biophys. Acta 1977, 491,
434–446.
[19] A. Schweiger, G. Jeschke, Principles of Pulse Electron Paramagnetic Reso-
nance, Oxford University Press, New York, 2001.
[20] K. Mꢄnnemann, D. Hinderberger, Physics, Chemistry, Biology and Medi-
cine: Spotlight on the Spin, Research Report 2011, MPI-P Mainz, Nr.
1179941.
Figure 10. Single-crystal structure of 2,2,6,6-tetramethylpiperidine-1-oxyl-4-yl
2,3,5-triiodobenzoate (TIB-SL).
[21] M. J. N. Junk, H. W. Spiess, D. Hinderberger, Angew. Chem. Int. Ed. 2010,
49, 8755–8759; Angew. Chem. 2010, 122, 8937–8941.
[22] M. J. N. Junk, H. W. Spiess, D. Hinderberger, J. Magn. Reson. 2011, 210,
210–217.
[23] B. J. Gaffney, in Spin Labeling: Theory and Applications: The Chemistry of
Spin Labels, Vol. 1 (Ed.: L. J. Berliner), Academic Press, New York, 1976,
pp. 183–232.
However, it should be noted that the single-crystal structure analy-
ses performed merely completed the characterization methods
listed in the Experimental Section for these newly synthesized SL
substances.
CCDC 1482421, 1482422, 1482423, and 1482424 contain the sup-
plementary crystallographic data for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic Data
Centre.
[24] J. B. Feix, C. S. Klug, in Biol. Magn. Reson.: Spin Labeling: The Next Millen-
nium, Vol. 14 (Ed.: L. J. Berliner), Plenum, New York, 1998, pp. 251–281.
[25] E. Bordignon, in Encyclopedia of Biophysics (S–Z): Spin-Labeling EPR of
Proteins: Dynamics and Water Accessibility of Spin-Label Side Chains, Vol.
5 (Ed.: G. C. K. Roberts), Springer, Heidelberg, 2013, pp. 2439–2447.
[26] J. P. Klare, H.-J. Steinhoff, Photosynth. Res. 2009, 102, 377–390.
[27] J. P. Klare, in Protein Interactions: Site-Directed Spin Labeling and Electron
Paramagnetic Resonance (EPR) Spectroscopy: A Versatile Tool to Study
Protein–Protein Interactions, Chap. 21 (Eds.: J. Cai and R. E. Wang),
InTech, Croatia, F, pp. 427–446.
[28] S. L. Kuan, B. Stçckle, J. Reichenwallner, D. Y. W. Ng, Y. Wu, M. Doroshen-
ko, K. Koynov, D. Hinderberger, K. Mꢄllen, T. Weil, Biomacromolecules
2013, 14, 367–376.
[29] Y. Akdogan, J. Reichenwallner, D. Hinderberger, PLoS ONE 2012, 7,
e45681.
Acknowledgements
We thank M. Beuchel (MPIP), S. Weber (MLU), H. Schimm
(MLU), C. Bauer (MPIP), and W. Scholdei (MPIP) for technical
support, and PD Dr. V. Mailꢃnder (MPIP) for help with all medi-
cal aspects. T.H. thanks S. Tꢄrk (MPIP) for help in FD mass spec-
trometry. This work was financially supported by the Deutsche
Forschungsgemeinschaft (DFG) under grant number HI1094/5-
1.
[30] S.-Y. Cheng, G. Rakhit, F. Erard, J. Robbins, C. F. Chignell, J. Biol. Chem.
1981, 256, 831–836.
[31] B. Neises, W. Steglich, Angew. Chem. Int. Ed. Engl. 1978, 17, 522–524;
Angew. Chem. 1978, 90, 556–557.
[32] S. van Doorslaer, G. Jeschke, in Fluxional Organometallic and Coordina-
tion Compounds: Dynamics by EPR: Picosecond to Microsecond Time
Scales, Chap. 6 (Eds.: M. Gielen, R. Willem, B. Wrackmeyer), Wiley, Chi-
chester, 2004, pp. 219–240.
Keywords: binding properties
spectroscopy · spin-labeling · structure–activity relationships
·
drug delivery
·
EPR
[33] G. Sudlow, D. J. Birkett, D. N. Wade, Mol. Pharmacol. 1975, 11, 824–832.
[34] G. Sudlow, D. J. Birkett, D. N. Wade, Mol. Pharmacol. 1976, 12, 1052–
1061.
[35] S. Stoll, A. Schweiger, J. Magn. Reson. 2006, 178, 42–55.
[36] D. J. Schneider, J. H. Freed, in Biol. Magn. Reson.: Spin Labeling: Theory
and Applications, Vol. 8 (Eds.: L. J. Berliner, J. Reuben), Plenum, New
York, 1989, pp. 1–76.
[1] S. Sugio, A. Kashima, S. Mochizuki, M. Noda, K. Kobayashi, Protein Eng.
1999, 12, 439–446.
[2] Th. Peters, Jr., in All about Albumin: Biochemistry Genetics, and Medical
Applications: 3: Ligand Binding by Albumin, Academic Press, San Diego,
1995, pp. 76–132.
[3] F. Yang, C. Bian, L. Zhu, G. Zhao, Z. Huang, M. Huang, J. Struct. Biol.
2007, 157, 348–355.
[37] G. Scatchard, Ann. N. Y. Acad. Sci. 1949, 51, 660–672.
[38] Lehninger Principles of Biochemistry: Box 12–1 Methods: Scatchard Anal-
ysis Quantifies the Receptor–Ligand Interaction (Eds.: D. L. Nelson, M. M.
Cox), 5th ed., W. H. Freeman, New York, 2008, p.421 ff.
[39] A. A. Spector, J. Lipid Res. 1975, 16, 165–179.
[4] X. M. He, D. C. Carter, Nature 1992, 358, 209–215.
[5] S. Leyon, J. F. Cramer, P. Nordberg, Acta Crystallogr. Sect. F 2008, 64, 64–
69.
[6] W. H. Pearlman, O. Crꢅpy, M. Murphy, J. Clin. Endocrinol. Metab. 1967,
27, 1012–1018.
[40] J. F. W. Keana, in Spin Labeling II: Theory and Applications: New Aspects
of Nitroxide Chemistry, Vol. 2 (Ed.: L. J. Berliner), Academic Press, New
York, 1979, pp. 115–166.
[41] G. I. Likhtenshtein, J. Yamauchi, S. Nakatsuji, A. I. Smirnow, R. Tamura, Ni-
troxides: Applications in Chemistry, Biomedicine and Materials Science
Wiley-VCH, Weinheim, 2008.
[7] J. Ghuman, P. A. Zunszain, I. Petitpas, A. A. Bhattacharya, M. Otagiri, S.
Curry, J. Mol. Biol. 2005, 353, 38–52.
[8] J. Reichenwallner, D. Hinderberger, Biochim. Biophys. Acta 2013, 1830,
5382–5393.
[9] N. Zaidi, E. Ahmad, M. Rehan, G. Rabbani, M. R. Ajmal, Y. Zaidi, N. Sub-
barao, R. H. Khan, J. Phys. Chem. B 2013, 117, 2595–2604.
[10] T. Chatterjee, A. Pal, S. Dey, B. K. Chatterjee, P. Chakrabarti, PLoS ONE
2012, 7, e37468.
[42] N. Kocherginsky and H. M. Swartz, Nitroxide Spin Labels: Reactions in
Biology and Chemistry, CRC, Boca Raton, 1995.
[11] S. S. Sinha, R. K. Mitra, S. K. Pal, J. Phys. Chem. B 2008, 112, 4884–4891.
[43] R. Dulbecco, M. Vogt, J. Exp. Med. 1954, 99, 167–182.
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
13
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
[44] J. L. Du, G. R. Eaton, S. S. Eaton, J. Magn. Reson. Ser. A 1995, 115, 213–
221.
Lottspeich, W. Sattler, M. Uhr, K. Mechtler, R. A. Dwek, P. M. Rudd, G.
Baier, H. Dieplinger, J. Proteome Res. 2005, 4, 889–899.
[52] H. J. Kayden, M. G. Traber, J. Lipid Res. 1993, 34, 343–358.
[53] A. A. Saboury, J. Iran. Chem. Soc. 2006, 3, 1–21.
[54] S. NfflÇez, J. Venhorst, C. G. Kruse, Drug Discovery Today 2012, 17, 10–
22.
[45] G. R. Desiraju and T. Steiner, The Weak Hydrogen Bond in Structural
Chemistry and Biology, Oxford University Press, Chichester, 1999.
[46] CRC Handbook of Tables for Organic Compound Identification, 3rd ed.,
compiled by Z. Rappoport, CRC, Cleveland, 1967.
[55] Y. Liang, Acta Biochim. Biophys. Sin. 2008, 40, 565–576.
[56] R. Ghai, R. J. Falconer, B. M. Collins, J. Mol. Recognit. 2012, 25, 32–52.
[57] H. Zhao, G. Piszczek, P. Schuck, Methods 2015, 76, 137–148.
[58] C. A. Brautigam, H. Zhao, C. Vargas, S. Keller, P. Schuck, Nat. Protoc.
2016, 11, 882–894.
[47] Springer Ser. Chem. Phys.: Spin Exchange: Principles and Applications in
Chemistry and Biology, Vol. 8 (Eds.: Y. N. Molin, K. M. Salikhov, K. I. Zamar-
aev), Springer, Berlin, 1980, ISBN 978-3-642-67668-0.
[48] R. Vieth, Clin. Chem. 1994, 40, 435–441.
[49] M. Kawakami, M. Imawari, D. S. Goodman, Biochem. J. 1979, 179, 413–
423.
[59] U. Kragh-Hansen, Dan. Med. Bull. 1990, 37, 57–84.
[50] B. W. Hollis, J. Steroid Biochem. 1984, 21, 81–86.
[51] L. Jerkovic, A. F. Voegele, S. Chwatal, F. Kronenberg, C. M. Radcliffe, M. R.
Wormald, E. M. Lobentanz, B. Ezeh, P. Eller, N. Dejori, B. Dieplinger, F.
Received: April 18, 2016
Published online on && &&, 0000
&
&
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
14
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
FULL PAPER
Find the link: Molecules from different
drug classes are labeled with EPR-active
nitroxide radicals (spin-labeled pharma-
ceuticals (SLPs)) and continuous-wave
(CW) EPR spectroscopy is used to inves-
tigate human serum albumin binding
under physiological conditions and at
varying ratios of SLP to protein (see
figure). Spectral simulations of the CW-
EPR spectra allow extraction of associa-
tion constants (KA) and the maximum
number (n) of binding sites per protein.
&
Structure–Activity Relationships
T. Hauenschild, J. Reichenwallner,
V. Enkelmann, D. Hinderberger*
&& – &&
Characterizing Active Pharmaceutical
Ingredient Binding to Human Serum
Albumin by Spin-Labeling and EPR
Spectroscopy
Chem. Eur. J. 2016, 22, 1 – 15
www.chemeurj.org
15
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ