Human serum albumin binding assay based on displacement of a non selective fluorescent inhibitor

Article abstract of DOI:10.1016/j.bmcl.2007.05.077

In this paper, we describe a fluorescent antibacterial analog, 6, with utility as a competition probe to determine affinities of other antibacterial analogs for human serum albumin (HSA). Analog 6 bound to HSA with an affinity of 400 ± 100 nM and the fluorescence was environmentally sensitive. With 370 nm excitation, environmental sensitivity was indicated by a quenching of the 530 nm emission when the probe bound to HSA. Displacement of dansylsarcosine from HSA by 6 indicated it competed with compounds that bound at site II (ibuprofen binding site) on HSA. Analog 6 also shifted the NMR peaks of an HSA bound oleic acid molecule that itself was affected by compounds that bound at site II. In addition to binding at site II, 6 interacted at site I (warfarin binding site) as indicated by displacement of dansylamide and the shifting of NMR peaks of an HSA bound oleic acid molecule affected by warfarin site binding. Additional evidence for multiple site interaction was discovered when a percentage of 6 could be displaced by either ibuprofen or phenylbutazone. A competition assay was established using 6 to determine relative affinities of other antibacterial inhibitors for HSA.

Full text of DOI:10.1016/j.bmcl.2007.05.077

Bioorganic & Medicinal Chemistry Letters 17 (2007) 4646–4649
Human serum albumin binding assay based on displacement
of a non selective fluorescent inhibitor
Atli Thorarensen,*, Ronald W. Sarver,*,à Fang Tian, Andrea Ho,
Donna L. Romero and Keith R. Marotti
Pharmacia Corporation, 301 Henrietta Street, Kalamazoo, MI 49001, USA
Received 27 February 2007; revised 23 May 2007; accepted 24 May 2007
Available online 27 May 2007
Abstract—In this paper, we describe a fluorescent antibacterial analog, 6, with utility as a competition probe to determine affinities
of other antibacterial analogs for human serum albumin (HSA). Analog 6 bound to HSA with an affinity of 400 100 nM and the
fluorescence was environmentally sensitive. With 370 nm excitation, environmental sensitivity was indicated by a quenching of the
530 nm emission when the probe bound to HSA. Displacement of dansylsarcosine from HSA by 6 indicated it competed with com-
pounds that bound at site II (ibuprofen binding site) on HSA. Analog 6 also shifted the NMR peaks of an HSA bound oleic acid
molecule that itself was affected by compounds that bound at site II. In addition to binding at site II, 6 interacted at site I (warfarin
binding site) as indicated by displacement of dansylamide and the shifting of NMR peaks of an HSA bound oleic acid molecule
affected by warfarin site binding. Additional evidence for multiple site interaction was discovered when a percentage of 6 could
be displaced by either ibuprofen or phenylbutazone. A competition assay was established using 6 to determine relative affinities
of other antibacterial inhibitors for HSA.
Ó 2007 Elsevier Ltd. All rights reserved.
We recently reported on the discovery of a novel class of
antibacterial agents, the anthranilates, which as a class
display potent broad-spectrum antibacterial activity.1
Extensive results from biological assays indicated that
high affinity to human serum albumin (HSA) is the main
reason for lack of in vivo activity for this series of com-
pounds.2 HSA is a 66.5 kDa monomeric protein com-
posed of three structurally similar a-helical domains, I
through III.3 These subdomains are further divided into
subdomain A containing six a-helices and subdomain B
containing four a-helices. Pioneering work on determin-
ing the binding sites of drugs indicated that high affinity
HSA binding of many drugs occurred at either of two
sites referred to as Sudlow sites I (warfarin site) and II
(ibuprofen site).4 These studies involved competitive dis-
placement of fluorescent probes from HSA by various
drugs and subsequent binding site classification based
on which probes were displaced. Two fluorescent probes
were particularly useful for site I and II classification.
Dansylsulfonamide (DNSA) bound to HSA with high
affinity at site I, Kd1 = 5.6 lM, and to another site with
much lower affinity Kd2 = 143 lM. Dansylsarcosine
bound to just site II with high affinity, Kd = 6 lM. Many
of the molecules investigated in the literature preferen-
tially bound at one of the two sites but iopanoic acid
bound to both sites I and II. Sudlow’s binding data also
indicated that probenecid, amitriptyline, and debrisoqu-
ine did not bind to either site I or II and they therefore
concluded there must be a third high affinity HSA bind-
ing site.
We evaluated numerous ways to measure the affinity of
our anthranilate analogs to HSA such as chromato-
graphic, spectroscopic, and the displacement of a high
affinity ligand from HSA.5 In addition we utilized the
addition of serum to the biological assay.2 The difference
of activity in the presence and absence of serum can be
related to HSA affinity.6 Of these techniques only the
addition of serum to our antibacterial assay and the
use of the Hummel–Dryer chromatography7 provided
us with data useful for guidance of the medicinal chem-
istry effort. The characterization of binding of 1 and 2
(Fig. 1) to HSA illustrated that the compounds bound
Keywords: Human serum albumin; Fluorescent probe.
*
Corresponding authors. Tel.: +1 636 247 3962; fax: +1 636 247
0250; e-mail addresses: atli.thorarensen@pfizer.com; Ronald.w.
sarver@pfizer.com
Present address: Pfizer, 700 Chesterfield, Parkway West AA2G,
Chesterfield, MO 63017, USA.
à
Present address: Pfizer, 2800 Plymouth Rd., Ann Arbor, MI 48105,
USA.
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.05.077

A. Thorarensen et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4646–4649
4647
than the absorption bands of most other translation
inhibitors, and an emission wavelength in a range with
Br
CO2H
NC
CO2H
A
B
little interference. Figure
2 shows the excitation
NH
O
NH
O
(Fig. 2a) and emission spectra (Fig. 2b) for 6. The exci-
tation maximum was at 370 nm and the emission maxi-
mum was at 533 nm. A linear response in intensity
versus concentration of 6 was detected over the range
examined.
O2S
N
O2S
N
2
1
O
Titration of 0.5 lM 6 with HSA (fraction V, Calbio-
chem catalog No. 126658) indicated the fluorescence
emission intensity of 6 was environmentally sensitive.
The fluorescence intensity of 6 at 533 nm decreased with
increasing concentration of HSA and there was a corre-
sponding increase in intensity at 470 nm (data not
shown). Fractional saturation of the HSA binding site(s)
for 6 can therefore be determined by following the
change in fluorescence of the probe as it binds or is
released from HSA. Figure 3 shows the average binding
isotherm obtained for the interaction of 6 with HSA.
The data represent the average isotherm from triplicate
titrations of 110 nM 6 with HSA in PBS at 23. °C. These
data were well fit using an equation for a single site bind-
ing isotherm resulting in an equilibrium dissociation
constant, Kd, of 0.39 lM. Although the data fit a single
site model, 6 could be displaced by compounds known
to bind at either site I or site II on HSA. Ibuprofen, war-
farin, diclofenac, and phenylbutazone were each able to
displace 6 from HSA. Ibuprofen has an affinity at site II
of ꢀ1 lM.3 Warfarin and phenylbutazone bind at site I
with Kd’s of 3 and 1.4 lM, respectively. High affinity
binding (Kd = 2 lM) of diclofenac was suggested to
occur at the benzodiazepine site with the second site
(Kd = 17 lM) shared with warfarin.12
Figure 1. Examples of characterized leads.
to multiple sites of HSA with high affinity. It was there-
fore of high desire to develop an HSA affinity assay that
could measure overall affinity of our compounds in a ra-
pid reliable manner.8 In this letter, we describe the dis-
covery of a fluorescent probe, its HSA binding profile,
and its utilization in a preliminary 96-well displacement
assay.
Identification of fluorescent probe. In a recent disclosure,
we discovered that the amide connecting the A and the B
rings could be replaced with a styrene linkage.9 This pro-
vided the avenue to prepare an extended conjugated sys-
tem. We prepared a series of heterocycles in this styrene
series both as potential antibacterial agents and as po-
tential fluorescent probes for HSA. The incorporation
of indole provided us with an analog 6, that was visibly
fluorescent during preparation and purification. The
conversion of indole acid 3 to the aldehyde was easily
accomplished (Scheme 1).10 A screen of numerous
reagents to stereospecifically prepare the E-alkene was
unsuccessful. The use of ylide 5 provided good yield of
the styrene but with no preference of olefin geometry.
Due to the inability to obtain selective olefination we
turned our attention to olefin isomerization utilizing thi-
ophenol and AIBN.11 The isomerization resulted in ca
12/1 (E/Z) mixture of olefins and now the desired isomer
6 could easily be recrystallized from MeOH in high yield
(Scheme 1).
In addition to the single concentration displacement
experiments, displacement of 6 from HSA was examined
HSA binding profile of 6. Ideally, a probe would have a
high quantum yield, an excitation wavelength higher
CN
1)
CO2Me
5
O
CO2H
N
1) LiAlH4
93%
PPh3
NaH
N
N
2) PCC
25%
2) PhSH/AIBN
RX E/Z 12.6/1
63%
3) NaOH
RX E/Z 1/0
75%
4
3
CO2H
NC
Figure 2. Excitation (a) and emission (b) spectra of 6 at 2.0 lM
(diamonds), 1.5 lM (triangles), 1.0 lM (squares), and 0.5 lM (circles)
in PBS. Additional spectra shown in each panel are baseline scan of
PBS.
6
Scheme 1.

4648
A. Thorarensen et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4646–4649
1.6
1.4
1.2
1.0
0.8
0.6
Data: Average1_Mean3Cuv
Model: SingleSiteHill
Equation: y=a/(1+(x/b)^c)+d
Weighting:
y
No weighting
Chi^2/DoF
= 0.00011
R^2
= 0.9991
DF
Kd
Hc
bkg
1.09162
0.39003
1.05513
0.53742
0.0185
0.01208
0.03603
0.01003
Figure 4. HSQC spectra of 13C-methyl labeled oleic acid complexed
with 0.5 mM HSA in 50 mM phosphate solution, pH 7.0. Mole ratio of
HSA: Oleic acid:6 = 1:4.2:1.0. Arrow A indicates the disappearance of
the resonance for oleic acid bound at the ibuprofen site. Arrow B
indicates the resonance for oleic acid bound at the warfarin site. This
resonance is altered when the mole ratio of analog 6 is increased to 2.0.
0.01
0.1
1
10
[HSA] μM
Figure 3. Data points are the average binding isotherm from triplicate
titrations of 110 nM 6 with HSA(V) in PBS, pH 7.0, error bars indicate
the standard deviations. Excitation was 370 nm with emission mon-
itored at 530 nm. The solid line resulted from least-squares curve fitting
of the data using the indicated equation for a single site binding
isotherm.
0
10
20
30
40
50
60
Time (min)
0.0
-0.2
-0.4
-0.6
-0.08
by titrating a solution of 6 bound to HSA with ibupro-
fen or phenylbutazone. The experiment had the starting
concentration of 6 at 0.5 lM with 19 lM HSA in PBS.
The large molar excess of HSA should ensure binding
of 6 at only the high affinity sites on HSA. First, phenyl-
butazone was titrated into the 6/HSA solution until little
change in fluorescence was detected. Then, ibuprofen
was titrated into that solution until the effect was satu-
rated. Ibuprofen was able to displace about 75 to 80%
of 6 bound to HSA and about 10% was displaced by
phenylbutazone.
-2
-4
-6
-8
Data: A021902A_NDH
Model: 1SITES
Chi^2 = 23889.95608
N
K
H
1.80
6.32E4 6.7E3
-1.052E4
0.056
532
Characterization of 6 binding to HSA by NMR. A recent
report from these laboratories has described an NMR
assay for characterization of ligand binding with
HSA.13 The binding of 6 to HSA was also characterized
by this NMR technique that uses the displacement of
oleic acid. In the presence of 6 at a molar ratio of
1.0:1.0 (HSA: 6), an oleic acid resonance from the ibu-
profen site was displaced (Figure 4, arrow A). While
the oleic acid resonance from the warfarin site was
shifted at a molar ratio of 1.0:2.0–2.0 (HSA: 6), oleic
acid resonance at arrow B was altered. Therefore, the
displacement of oleic acid experiment suggests that 6
can effect binding of oleic acids at both the ibuprofen
and warfarin sites.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Figure 5. Isothermal titrating calorimetry results for 6 interaction with
HSA (V) in PBS. The top plot shows the enthalpy change per 6 lL
injection of a 0.5 mM solution of 6 into a 20 lM solution of HSA (V)
in PBS at 35 °C. Bottom panel, data points are the integrated
enthalpies for each injection and the solid line results from non linear
least-squares fitting of the data to the equation for a single site binding
model.14
molecules of 6 bound per HSA. The enthalpy change
for the interaction was À10.4 kcal/M with an average
affinity of 15 lM.
Characterization of 6 binding to HSA by ITC. Figure 5
shows the binding isotherm for the interaction of 6 with
HSA as measured by isothermal titrating calorimetry
(ITC). The top panel shows the enthalpy change per
6 lL injection of 6 into 20 lM HSA. The heat of dilu-
tion of 6 into buffer was negligible. The bottom panel
of Figure 5 shows the integrated enthalpy change for
each injection and the best fit line using a single site
model. The agreement between the model and the exper-
imental data was good and indicated approximately 2
A difference in the Kd determined by ITC compared to
the fluorescence method was not surprising given the
complexity of the binding interaction. The ITC tech-
nique was performed by titrating 6 into HSA until an
excess of 6 saturated the binding interaction, while the
fluorescence method titrated 6 with HSA until excess
HSA saturated the effect. With ITC, the response is
due to binding of 6 to HSA at a high affinity site or sites.
As these become saturated, lower affinity sites start to
fill and the enthalpy change for these interactions is also

A. Thorarensen et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4646–4649
4649
bound at site I or II on HSA competed with 6 for
HSA binding. Analog 6 was useful for determining
affinities of other antibacterial analogs and helped
guide syntheses of compounds with reduced HSA affin-
ity. Good correlations between HSA affinity deter-
mined using 6 and the effect of serum on minimum
inhibitor concentrations (MICs) were found for several
sub-classes of antibacterial compounds. The assay per-
formed in 96-well plates has relatively good sample
throughput, ꢀ100 compounds a day in our laboratory
and could easily be expanded to a higher density
format.
References and notes
1. Larsen, S. D.; Hester, M. R.; Ruble, J. C.; Kamilar, G.
M.; Romero, D. L.; Wakefield, B.; Melchior, E. P.;
Sweeney, M. T.; Marotti, K. R. Bioorg. Med. Chem. Lett.
2006, 6173.
2. Li, J.; Wakefield, B. D.; Ruble, J. C.; Stiff, C. M.; Romero,
D. L.; Marotti, K. R.; Sweeney, M. T.; Zurenko, G. E.;
Rohrer, D. C.; Thorarensen, A. Bioorg. Med. Chem. Lett.
2007, 2347.
Figure 6. The graph shows the ratio of the antibacterial activity of the
series of thioether compounds utilizing the MIC values (vs S. aureus
9218) for each compound with and without 10% serum in relation to
the displacement of that compound by 6 from HSA.
3. Peters, T., Jr. All About Albumin, Biochemistry, Genetics,
and Medical Applications; Academic: New York, 1996,
Chapter 3.
4. Sudlow, G.; Birkett, D. J.; Wade, D. N. Mol. Pharm. 1975,
11, 824.
5. (a) Epps, D. E.; Raub, T. J.; Caiolfa, V.; Chiari, A.;
Zamai, M. J. Pharm. Pharmacol. 1999, 51, 41; (b) Epps, D.
E.; Raub, T. J.; Kezdy, F. J. Anal. Biochem. 1995, 227,
detected. If binding enthalpies for the various sites are
not sufficiently different, the measured enthalpy change
is a convolution of enthalpy at each site and no sharp
break is detected in the isotherm at set stoichiometries.
Such was the case for 6. With the fluorescence method,
HSA is in excess and 6 binding occurs mainly at high
affinity sites.
´
342; (c) Oravcova, J.; Bo¨hs, B.; Lindner, W. J. Chroma-
togr. B 1996, 677, 1.
Affinity determined by competitive displacement of 6.
Initially, several translation inhibitors with thioether
appendages were examined using the competitive
displacement assay.8 The percentage displacement of
0.5 lM 6 from 1.4 lM HSA for each of the compounds
was tested at 12 lM. Larger percentage displacement
indicates tighter relative HSA affinity for the compound.
Kd’s assuming a single site binding model for 6 to HSA
were calculated but were considered approximate since
6 binds to more than one site. This was compared to
our alternative way of measuring protein binding which
was addition of serum to the antibacterial assay, Figure
6. There appears to be a reasonable relationship between
the two assay formats but with this new assay the benefit is
the direct measurement of HSA affinity irrespective of the
antibacterial activity of the compound. This assay was
subsequently utilized to screen hundreds of compounds
guiding the medicinal chemistry effort going forward.
6. Copeland, R. A. J. Pharm. Sci. 2000, 89, 1000.
7. Rogers, J. M.; DeZwaan, J. Unpublished results.
8. (a) Thorarensen, A.; Li, J.; Wakefield, B. D.; Romero, D.
L.; Marotti, K. M.; Sweeney, M. T.; Zurenko, G.; Sarver,
R. W. Bioorg. Med. Chem. Lett. 2007, 3113; (b) Stiff, C.
M.; Zhong, M.; Sarver, R. W.; Gao, H.; Ho, A.; Sweeney,
M. T.; Zurenko, G.; Romero, D. L. Bioorg. Med. Chem.
Lett. 2007, doi:10.1016/j.bmcl.2007.05.006.
9. Thorarensen, A.; Wakefield, B. D.; Romero, D. L.;
Marotti, K. M.; Sweeney, M. T.; Zurenko, G.; Rohrer,
D. C.; Han, F.; Bryant, G. L., Jr. Bioorg. Med. Chem.
Lett. 2007, 2823.
10. Kasmai, H. S.; Mischke, S. G.; Blake, T. J. J. Org. Chem.
1995, 60, 2267.
11. Schwarz, M.; Graminski, G. F.; Waters, R. M. J. Org.
Chem. 1986, 51, 260.
12. Chamouard, J.; Barre, J.; Urien, S.; Houin, G.; Tillement,
P. Biochem. Pharm. 1985, 34, 1695.
13. Sarver, R. W.; Gao, H.; Tian, F. Anal. Biochem. 2005, 347,
297.
Conclusion. An HSA affinity assay was developed using
a fluorescent antibacterial analog 6. Compounds that
14. Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L. N. Anal.
Biochem. 1989, 179, 131.