Hydrophilic molecular rotor derivatives - Synthesis and characterization

Article abstract of DOI:10.1016/j.bioorg.2004.04.002

Recent research shows high potential for some p-N,N- dialkylaminobenzylidenecyanoacetates, part of a group known as fluorescent molecular rotors, to serve as fluorescent, non-mechanical viscosity sensors. Of particular interest are molecules compatible with aqueous environments. In this study, we present the synthesis and physical characterization of derivatives from 9-(2-carboxy-2-cyanovinyl)-julolidine and related molecules. All compounds show a power-law relationship of fluorescence emission with the viscosity of the solvent, different mixtures of ethylene glycol and glycerol to modulate viscosity. Compounds with high water solubility exhibit the same behavior in aqueous solutions of dextran, where the dextran concentration was varied to modulate viscosity. In addition, some compounds have been found to have low sensitivity towards changes in the pH in the physiological range. The compounds presented show promise to be used in biofluids, such as blood plasma or lymphatic fluid, to rapidly and non-mechanically determine viscosity.

Full text of DOI:10.1016/j.bioorg.2004.04.002

Bioorganic Chemistry 32 (2004) 274–289
www.elsevier.com/locate/bioorg
Hydrophilic molecular rotor
derivatives—synthesis and characterization
Mark A. Haidekker,a,¤ Thomas P. Brady,b Sevag H. Chalian,b
Walter Akers,a Darcy Lichlyter,a
and Emmanuel A. Theodorakisb
a Department of Biological Engineering, University of Missouri-Columbia, Columbia, MO, USA
b Department of Chemistry and Biochemistry, University of California, San Diego, USA
Received 22 March 2004
Available online 13 May 2004
Abstract
Recent research shows high potential for some p-N,N-dialkylaminobenzylidenecyanoace-
tates, part of a group known as Xuorescent molecular rotors, to serve as Xuorescent, non-
mechanical viscosity sensors. Of particular interest are molecules compatible with aqueous
environments. In this study, we present the synthesis and physical characterization of deriva-
tives from 9-(2-carboxy-2-cyanovinyl)-julolidine and related molecules. All compounds show a
power-law relationship of Xuorescence emission with the viscosity of the solvent, diVerent mix-
tures of ethylene glycol and glycerol to modulate viscosity. Compounds with high water solu-
bility exhibit the same behavior in aqueous solutions of dextran, where the dextran
concentration was varied to modulate viscosity. In addition, some compounds have been
found to have low sensitivity towards changes in the pH in the physiological range. The com-
pounds presented show promise to be used in bioXuids, such as blood plasma or lymphatic
Xuid, to rapidly and non-mechanically determine viscosity.
 2004 Elsevier Inc. All rights reserved.
1. Introduction
In biological systems, viscosity plays an important role on scales from the cellular
level to the organism level. For example, cytoplasmic viscosity plays a role in cell
¤ Corresponding author. Fax: 1-573-884-5650.
E-mail address: HaidekkerM@missouri.edu (M.A. Haidekker).
0045-2068/$ - see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2004.04.002

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
275
motion and speciWc lung diseases [1,2]. Cell membrane viscosity has been linked to
cell mechanosensing [1,3,4], the action of anaesthetics and alcohol [5–8], and a range
of diseases including atherosclerosis, cell malignancy, hypercholesterolemia, and dia-
betes [9–13]. On a more macroscopic level, the viscosity of body Xuids, including
blood plasma [14–16] and lymphatic Xuid [17,18], has been linked to various disease
states. In addition, the measurement and management of the appropriate blood
plasma viscosity is becoming an important issue in transfusion medicine, where new
strategies seeking the development of artiWcial blood are pursued [19–21].
The use of mechanical forces has been the primary method of measuring viscosity
levels in bioXuids. One type of mechanical viscometer is the cone-and-plate viscometer,
where Xuid is sheared between a Xat plate and a cone. In this way, the resistance of the
Xuid to this force is measured. This is not only a time-consuming method, requiring
thorough cleaning of the instrument between each measurement, but results are also
inXuenced by artefacts, such as the formation of a protein layer at the Xuid–air inter-
face, leading to apparent non-Newtonian properties [14,22]. These drawbacks may be
one reason for the limited use of viscosity measurement techniques in a clinical setting.
To overcome these limitations an alternative strategy for the measurement of
viscosity has been developed based on the use of viscosity-sensitive Xuorescent mole-
cules. These compounds, commonly referred to as molecular rotors, show a well-
deWned relationship between Xuorescence quantum yield and solvent viscosity [23].
Thus, determination of viscosity changes can be reduced to the simple measurement
of Xuorescence intensity [4,24–26]. In addition, Xuorescence can be acquired at
microscopic levels, allowing for monitoring of microviscosity [24].
For the past several years we have been engaged in the development of molecular
rotors for probing the microviscosity in the lipophilic regions of the cell membrane
[4,27,28]. These probes are unsuitable for use in aqueous environments due to their
poor solubility. With the growing importance of viscosity measurements in aqueous
solutions, we now present a group of molecular rotors that exhibit good solubility in
water and can therefore be used as viscosity probes in aqueous colloid solutions and
bioXuids.
2. Materials and methods
2.1. General techniques
Compounds 1 (DCVJ) and 6 (CCVJ) were purchased from Helix Research
(SpringWeld, OR). Aldehyde 7 and cyanoacetic acid methyl ester (5e) were purchased
from Aldrich. Aldehyde 4 and compound 2e were synthesized as previously described
[28]. All reagents were commercially obtained at highest commercial quality and used
without further puriWcation. Diethyl ether (Et2O), dichloromethane (CH2Cl2), tolu-
ene (PhCH3), and benzene (PhH) were puriWed by passage through a bed of activated
alumina. Yields refer to chromatographically and spectroscopically (1H NMR, 13C
NMR) homogeneous materials, unless otherwise stated. Reactions were monitored
by thin-layer chromatography (TLC) carried out on 0.25mm E. Merck silica gel

276
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
plates (60F-254) using UV light as the visualizing agent and 10% ethanolic phospho-
molybdic acid (PMA) or p-anisaldehyde solution and heat as developing agents.
E. Merck silica gel (60, particle size 0.040–0.063 mm) was used for Xash chromatogra-
phy. Preparative thin-layer chromatography separations were carried out on 0.25 or
0.50 mm E. Merck silica gel plates (60F-254). NMR spectra were recorded on Varian
Mercury 400 and/or Unity 500MHz instruments and calibrated using the residual
undeuterated solvent as an internal reference. The following abbreviations were used
to explain the multiplicities: sD singlet, dD doublet, tDtriplet, qD quartet, m D
multiplet, and b Dbroad. IR spectra were recorded on a Nicolet 320 Avatar FT-IR
spectrometer and values are reported in cm¡1 units. Optical rotations were
recorded on a Jasco P-1010 polarimeter and values are reported as follows: [ꢀ]Tꢁ (c:
g/100ml, solvent). High resolution mass spectra (HRMS) were recorded on a VG
7070 HS mass spectrometer under chemical ionization (CI) conditions or on a VG
ZAB-ZSE mass spectrometer under fast atom bombardment (FAB) conditions.
2.2. General procedure for the preparation of ꢀ-cyanoesters 5a, 5c, 5d, and 8
To a solution of cyano acetic acid (5) (10 mmol), the corresponding alcohol
(10mmol) in dichloromethane (20 ml), and DMAP (1mmol) at 0 °C was added drop-
wise a solution of DCC (10 mmol) in dichloromethane (10ml). After the addition was
completed, the mixture was allowed to warm to 25°C where it was stirred for 2 h. The
reaction was then diluted with dichloromethane (20 ml) and the formed DCU was
Wltered under gravity. The Wltrate was dried (MgSO4) and concentrated and the resi-
due was puriWed by column chromatography (silica, 5–20% ethyl ether in hexane) to
aVord the corresponding esters.
2.2.1. Ester 5a
78% yield; colorless liquid; Rf D 0.5 (50% ether in hexanes); IR (Wlm) ꢁmax 2841–
1
2905, 2361, 1732, 1515, 1465, 1384, 1338, 1248, 1171cm¡1; H NMR (400 MHz,
CDCl3) ꢂ 7.29 (d, J D8.4Hz, 2H), 6.89 (d, J D8.4Hz, 2H), 5.04 (s, 2H), 4.12
(t, J D 6.8 Hz, 2H), 3.81 (s, 3H), 3.45 (s, 2H), 2.31 (t, J D 6.8 Hz, 2H), 1.56 (m, 6H), 1.24
(s, 20H); 13C NMR (100 MHz, CDCl3) ꢂ 173.8, 162.9, 159.4, 130.0, 128.2, 113.9, 113.0,
67.1, 65.9, 55.2, 34.3, 32.8, 29.6, 29.5, 29.4, 29.2, 29.1, 28.3, 25.6, 24.9; HRMS: Calcd
for C27H41NO5: (M + H+) 460.3063. Found 460.3089.
2.2.2. Ester 5c
90% yield; colorless liquid; Rf D0.5 (100% ether); IR (Wlm) ꢁmax 2886, 2252, 1746
1
1451, 1388, 1334, 1252, 1193, 1103cm¡1; H NMR (400MHz, CDCl3) ꢂ 4.28 (m,
2H), 3.66 (m, 2H), 3.58 (m, 2H), 3.53 (m, 2H), 3.49 (s, 2H), 3.46 (q, JD7.2Hz, 2H), 1.30
(t, JD7.2Hz, 3H); 13C NMR (100MHz, CDCl3) ꢂ 163.0, 113.0, 70.2, 69.3, 68.2, 66.4,
65.4, 24.4, 14.7; HRMS: Calcd for C9H15NO4: (M+H+) 202.1079. Found 202.1091.
2.2.3. Ester 5d
86% yield; Rf D0.4 (100% ether); IR (Wlm) ꢁmax 2886, 2252, 1746, 1451, 1388, 1334,
1252, 1193, 1103cm¡1; 1H NMR (400 MHz, CDCl3) ꢂ 4.29 (m, 2H), 3.67 (m, 2H), 3.59

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
277
(m, 6H), 3.50 (m, 2H), 3.49 (s, 2H), 3.32 (s, 3H); 13C NMR (100 MHz, CDCl3) ꢂ 163.0,
113.0, 71.7, 70.4, 70.3, 68.3, 65.5, 58.8, 24.5; HRMS: Calcd for C10H17NO5: (M + H+)
232.1185. Found 232.1199.
2.2.4. Ester 8
74% yield; colorless liquid; Rf D0.5 (90% ether in hexanes); IR (Wlm) ꢁmax 2932–
1
2977, 2262, 1750, 1452, 1375, 1334, 1257, 1193, 1049cm¡1; H NMR (400 MHz,
CDCl3) ꢂ 4.35 (m, J D4.4 Hz, 1H), 4.22 (td, J D 4.4 Hz, J D11.2Hz, J D 26.4 Hz, 2H),
4.07 (dd, J D6.4Hz, J D6.8 Hz, 1H), 3.73 (dd, J D5.6Hz, J D 6.0 Hz, 1H), 3.51 (s, 2H),
1.40 (s, 3H), 1.33 (s, 3H); 13C NMR (100MHz, CDCl3) ꢂ 162.8, 112.8, 110.0, 109.2,
76.0, 72.9, 66.6, 65.8, 26.6, 25.1; HRMS: Calcd for C9H13NO4: (M + H+) 200.0923.
Found 200.0931.
2.3. General procedure for the preparation of compounds 2a, 2c, 2d, 10, and 11
To a solution of 4-formyl julolidine (4) (2 mmol) and ester 5 (2mmol) in THF
(10ml) was added DBU (2.1 mmol). After stirring at 25 °C for 1 h, the mixture was
concentrated and the residue subjected to silica gel chromatography (10–50% ether in
hexanes) to aVord the corresponding adduct 2.
2.3.1. Ester 2a
85% yield; bright orange liquid. Immediately after isolation this compound was
treated with a solution of TFA/anisole (95/5) for 15 min at 25°C. The TFA was
removed under reduced pressure and the residue puriWed using silica gel chromatog-
raphy to aVord carboxylic acid 2b (71% yield). Ester 2b: Rf D0.5 (50% ether in hex-
1
anes); IR (Wlm) ꢁmax 2850–2923, 1701, 1560, 1519, 1316, 1230, 1162cm¡1; H NMR
(400 MHz, CDCl3) ꢂ 7.94 (s, 1H), 7.51 (s, 2H), 4.25 (t, J D6.4 Hz, 2H), 3.34
(t, J D 6.0 Hz, 4H), 2.75 (t, J D6.8 Hz, 4H), 2.35 (t, J D 7.6 Hz, 2H), 1.96 (m, 4H), 1.75
(m, J D 6.8 Hz, 2H), 1.64 (m, J D7.2 Hz, 2H), 1.26 (m, 22H); 13C NMR (100 MHz,
CDCl3) ꢂ 179.5, 164.9, 154.3, 147.6, 131.6, 120.7, 118.4, 118.0, 91.6, 65.8, 50.1, 34.0,
29.6, 29.5, 29.4, 29.2, 29.0, 28.6, 27.5, 25.8, 24.7, 21.1; HRMS: Calcd for C32H46N2O4:
(M+ H+) 523.3536. Found 523.3554.
2.3.2. Ester 2c
78% yield; bright red oil; Rf D 0.7 (100% ether); IR (Wlm) ꢁmax 2859–2931, 2207,
1
1705, 1614, 1565, 1519, 1442, 1316, 1230, 1161, 1098cm¡1; H NMR (400 MHz,
CDCl3) ꢂ 7.88 (s, 1H), 7.45 (s, 2H), 4.36 (t, J D 4.8 Hz, 2H), 3.77 (t, J D 3.6Hz, 2H), 3.68
(m, 2H), 3.56 (m, 2H), 3.49 (m, 2H), 3.29 (t, J D 6 Hz, 4H), 2.70 (t, J D6Hz, 4H), 1.91
(m, J D 6 Hz, 4H), 1.17 (t, J D 6.6 Hz, 3H); 13C NMR (100MHz, CDCl3) ꢂ 164.6, 154.3,
147.6, 131.5, 120.6, 118.1, 90.9, 70.7, 69.8, 68.8, 66.5, 64.6, 50.0, 27.4, 20.9, 15.0; HRMS:
Calcd for C22H28N2O4: (M + H+) 385.2127. Found 385.2141.
2.3.3. Ester 2d
84% yield; bright red oil; Rf D 0.6 (100% ether); IR (Wlm) ꢁmax 2859–2941, 2352,
2198, 1705, 1610, 1560, 1519, 1442, 1316, 1225, 1162, 1098cm¡1; 1H NMR (400 MHz,

278
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
CDCl3) ꢂ 7.92 (s, 1H), 7.49 (s, 2H), 4.39 (t, J D4.8 Hz, 2H), 3.79 (t, J D 5.2 Hz, 2H), 3.72
(m, 2H), 3.66 (m, 4H), 3.55 (m, 2H), 3.37 (s, 3H), 3.32 (t, J D8.8 Hz, 4H), 2.73 (t,
J D6.4 Hz, 4H), 1.95 (m, J D 5.6 Hz 4H); 13C NMR (100MHz, CDCl3) ꢂ 164.7, 154.5,
147.7, 131.7, 120.8, 118.4, 91.3, 71.9, 70.8, 70.6, 70.5, 68.9, 64.8, 50.1, 27.5, 21.0; HRMS:
Calcd for C23H30N2O5: (M + H+) 415.2233. Found 415.2259.
2.3.4. Acetonide 9
69% yield; Rf D0.4 (70% ethyl ether); IR (Wlm) ꢁmax 2937, 2207, 1703, 1619, 1565,
1513, 1439, 1311, 1229, 1157, 1098cm¡1; 1H NMR (400MHz, CDCl3) ꢂ 8.08 (s, 1H),
7.92 (d, J D8.8Hz, 2H), 6.70 (d, J D8.8 Hz, 2H), 4.40 (m, J D 6Hz, 1H), 4.32 (m, 2H),
4.13 (dd, J D 6, 8.4 Hz, 1H), 3.88 (dd, J D5.6, 8.4 Hz, 1H), 3.11 (s, 6H), 1.46 (s, 3H), 1.38
(s, 3H); 13C NMR (100MHz, CDCl3) ꢂ 164.1, 155.0, 153.7, 134.2, 119.3, 117.2, 111.5,
109.8, 93.2, 73.4, 66.4, 65.4, 40.0, 26.7, 25.5; HRMS: Calcd for C18H22N2O4: (M + H+)
331.1658. Found 331.1691.
2.3.5. Diol 10
Compound 9 (0.5g, 1.5mmol) was dissolved in a mixture of THF/MeOH (1/1) and
DOWEX-H+ resin (0.10g) was added and heterogeneous mixture was stirred for 2h.
The DOWEX-H+ resin was removed by Wltration and triethylamine (50mg, 0.5mmol)
was added and the solvent was removed on the rotary evaporator. The residue was
puriWed on silica gel (100% ether) to aVord 200mg of compound 10 as a bright yellow
solid (46%). 10: Rf D0.3 (100% ethyl ether); IR (Wlm) ꢁmax 3412, 2932, 2207, 1701, 1610,
1565, 1528, 1451, 1379, 1329, 1274, 1225, 1184cm¡1; 1H NMR (400MHz, CDCl3) ꢂ 8.08
(s, 1H), 7.94 (d, JD9.2Hz, 2H), 6.70 (d, JD9.2Hz, 2H), 4.37 (dt, JD4.4, 11.2, 23.6Hz,
2H), 4.06 (m, JD4.8Hz, 1H), 3.74 (dt, JD4, 12, 29.6Hz, 2H), 3.13 (s, 6H); 13C NMR
(100MHz, CDCl3) ꢂ 164.7, 155.1, 153.8, 134.2, 119.2, 117.6, 111.5, 92.5, 70.0, 66.7, 63.3,
40.1; HRMS: Calcd for C15H18N2O4: (M+H+) 291.1345. Found 291.1361.
2.3.6. Compound 11
Compound 11 was prepared from condensation of 4-dimethylamino-benzalde-
hyde (7) (1g, 6.7 mmol), cyano acetic acid methyl ester (0.8 g, 6.7 mmol), and DBU
(1.0g, 6.7mmol) in 15ml of THF as described for compound 5d. Isolated yield: 1.4 g,
93%. 11: Rf D0.4 (50% ether in hexanes). IR (Wlm) ꢁmax 2937, 1703, 1619, 1565, 1513,
1439, 1157, 1098 cm¡1; 1H NMR (400 MHz, CDCl3) ꢂ 8.07 (s, 1H), 7.93 (d, J D 8.8Hz
2H), 6.68 (d, J D8.8 Hz, 2H), 3.88 (s, 3H), 3.10 (s, 6H); 13C NMR (100 MHz, CDCl3) ꢂ
164.8, 154.7, 153.6, 134.1, 119.2, 117.6, 111.4, 93.3, 52.8, 39.9; HRMS: Calcd for
C13H14N2O2: (M + H+) 231.1133. Found 231.1156.
2.4. Spectral properties
Reagents were purchased as follows: Xuoroscopy grade DMSO and dextran
(87kDa average MW) from Sigma (St. Louis, MO), ethylene glycol and glycerol from
Fisher ScientiWc (Pittsburgh, PA). Fluorescence spectroscopy were performed in meth-
ylacrylate microcuvettes (Fisher ScientiWc) using Spex Fluoromax-3 (Jobin-Yvon,
Stanmore, North London, UK). Based on dye solubility, a 20, 8.87 or 5mM stock

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
279
solution in DMSO was prepared from each of the dyes listed and further diluted as
needed: To determine quantum yield, the dyes were diluted to 5ꢂM in ethylene glycol.
Mixtures of ethylene glycol (EG) and glycerol were prepared by pre-staining 5ml of
EG with 100ꢂl dye stock solution. One milliliter of pre-stained EG was then mixed
with each 1ml unstained EG and 8ml glycerol, with 2ml EG and 7 ml glycerol, with
3ml EG and 6ml glycerol, with 4ml EG and 5ml glycerol, and with 7 ml EG and 2ml
glycerol to aVord solutions of diVerent viscosity and constant dye concentration.
Since the spectral properties of each molecular rotor diVers from the others, and
the spectral maxima are to a small extent aVected by solution polarity, the excitation
and emission wavelength was examined for each dye before every experiment and
optimized accordingly. Typically, 1 nm increment with 0.3 s integration and 5 nm
bandwidth for excitation and emission slits was used. Peak emission intensity was
calculated by averaging Wve data points (5 nm) centered on the peak emission.
2.5. Aqueous solutions
The aqueous solutions consisted of phosphate-buVered saline (PBS) from Sigma
(St. Louis, MO) pH 7.4, with 10 ꢂM of each dye tested. The dextran was added to
make the following weight/volume solutions, 2.5, 5.0, 7.5, and 10.0%. Acid/base solu-
tions used during pH assays consisted of distilled water; the pH was adjusted with
HCl or NaOH. All solutions were calibrated and adjusted for pH before every exper-
iment with an Oakion pH meter for accuracy.
2.6. Viscosity measurement
Viscosity was measured using a computer-controlled BrookWeld DV-III+ rheome-
ter. The viscosity was measured at ten speed settings which were determined by the
percent torque on the spindle (maintained between 10 and 100%). Measurements
started at 10% torque (low rpm) and were stepped up in four regular intervals to 90%
torque, and measurements were subsequently repeated while stepping back down.
The viscosity measurement utilized in the data is the middle step, generally 30–60%
torque, as the speed was decreased back to 10% torque. Five independent measure-
ments were taken for each solution tested and the results averaged.
3. Results
3.1. Synthesis of the molecular rotors
The general schematic of the molecular rotors analyzed can be seen in Fig. 1. The
synthesis of compounds 2a–2e and 10 are shown in Schemes 1 and 2, respectively.
Aldehyde 4 and compound 2e were synthesized as previously described [28]. The
ꢀ-cyanoesters 5a, 5c, 5d, and 8 were obtained via a dicyclohexyl carbodiimide-induced
esteriWcation of cyanoacetic acid with the corresponding alcohols in 80–95% yield.
Condensation of ꢀ-cyanoesters 5 with aldehyde 4 in the presence of DBU produced

280
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
Fig. 1. General structure of the hydrophilic rotors examined (2 and 3) in comparison to the established dye
DCVJ (1).
Scheme 1. Synthesis of compounds 2a–2e.
the desired probes 2a, 2c–2e in very good yields (74–90%). Ester 2a was converted to
carboxylic acid 2b upon treatment with TFA in anisole in 89% yield. In a similar
manner, condensation of aldehyde 4 with ester 5e produced compound 11. Conden-
sation of 4 with 8 produced compound 9 that after treatment with 1N HCl gave rise
to diol 10. All compounds were puriWed by chromatography on silica gel prior to use.
3.2. Spectral properties
All molecules examined exhibited Xuorescence in a similar spectral range as CCVJ.
Generally, absorption was found in the blue range in the area of 420–
460nm, and emission was in the range 470–510 nm (green). Spectral maxima for
all dyes are listed in Table 1. Representative excitation and emission spectra for dyes
2d, 6, 10, and 11 can be seen in Fig. 2. A comparison of quantum yields was per-
formed by measuring emission intensity and computing intensity relative to emission

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
281
Scheme 2. Synthesis of compound 10 and structure of compound 11.
intensity of FITC under identical conditions. In order to obtain acceptable emission
signals, quantum yield comparison was performed in ethylene glycol. The relatively
high viscosity of ethylene glycol elevates the emission intensity approximately by a
factor of 5 over water. Table 1 lists the relative intensities of all dyes.
3.3. Water solubility
In order to obtain a qualitative evaluation of water solubility, 10ꢂM of each dye
was dissolved in water, and the solution was tested for precipitates under the epiXuo-
rescent microscope. Dyes with high solubility showed a lower number of precipitated
crystals than dyes with lower solubility, such as DCVJ. All dyes tested, including 11,
dissolved in water after less than 1min of stirring, and did not show precipitates. The
exception was 2b, which showed few, but very large Xuorescent clusters after a short
time. This indicates that 2b had limited solubility, but aggregated after a short time
Table 1
Spectral properties of the dyes analyzed
Dye
Excitation maximum
(nm)
Emission maximum
(nm)
Relative quantum yield
in % of FITC
2b
2c
2d
10
459
461
460
442
440
460
420
494
502
502
503
485
486
502
486
522
36.0
35.0
28.4
35.7
27.2
27.7
33.8
100
11
1a
6a
FITCa
a Established dyes presented for comparison. Note that FITC is not a molecular rotor.

282
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
Fig. 2. Excitation and emission spectra of selected molecular rotors (2d, 10, and 11) in comparison to the
established dye CCVJ (6).
and fell out of solution. In addition, 2b exhibited rapid Xuorescence decay (photoble-
aching) under the microscope.
3.4. Viscosity sensitivity
All dyes examined in this study exhibited an increase of emission intensity in solu-
tions of increased viscosity. To determine the individual dye's viscosity sensitivity,
mixtures of ethylene glycol and glycerol were used to modulate solvent viscosity. By
plotting the maximum spectral intensity over the solvent viscosity in a double-loga-
rithmic scale, it was possible to determine the power-law relationship between inten-
sity and viscosity. Representative curves for dyes 2d, 6, 10, and 11 can be seen in
Fig. 3, and Table 2 lists the slopes together with the regression parameters for all
dyes.
Viscosity-sensing performance was also tested in aqueous solutions of a high
molecular-weight colloid, dextran. Viscosity sensitivity of 6 in such aqueous colloid
Fig. 3. Graphs showing the peak emission intensity over the solvent viscosity in a double-logarithmic scale
for selected molecular rotors (2d, 10, and 11) in comparison to the established dye CCVJ (6). It can be seen
that the values lie on a straight line, conWrming the power-law relationship of intensity with viscosity.

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
283
Table 2
Regression results of log intensity over log viscosity to determine the exponent x in Eq. (2)
Dye
Regression slope
r2
P
(exponent x)
2b
2c
2d
10
11
6a
0.3
0.81
0.99
0.99
0.99
0.77
0.98
0.036
0.61
0.59
0.54
0.32
0.52
00.0001
00.0001
0.0002
00.05
0.0014
a Established dye presented for comparison.
Fig. 4. Graphs showing the peak emission intensity in aqueous solutions of dextran at diVerent concentra-
tions. Shown is the intensity over the solvent viscosity in a double-logarithmic scale for selected molecular
rotors (2d, 10, and 11). Similar to the behavior in ethylene glycol–glycerol mixtures, these probes exhibit
power-law relationship between intensity and viscosity in aqueous colloid solutions.
systems has been established recently [29]. The new water-soluble dyes, 2d, 10, and 11,
underwent testing in water/dextran solutions. Fig. 4 shows the intensity increase with
increasing concentration of dextran. The power-law relationship can again clearly be
seen. The slopes in the double-logarithmic plot were found to be 0.44 (r2 D 0.99, 2d),
0.37 (r2 D0.99, 10), and 0.44 (r2 D 0.99, 11). While the slopes were signiWcantly diVer-
ent from zero in all cases with P 0 0.0001, they were markedly lower than in the
ethylene glycol/glycerol mix.
3.5. Sensitivity towards changes in the pH of the environment
Since these dyes are targeted at aqueous solutions, possibly at biological Xuids
with varying acidity, the inXuence of the solvent pH on the emission intensity was
examined. As can be seen in Fig. 5, the dyes examined showed little variation of the
emission intensity with pH in the pH range from 4.0 to 12.0. In the pH range from 3.0
to 9.0, 2d shows no correlation (r2 D 0.05, P D 0.3, not signiWcant) between pH and
intensity, while a slight positive correlation was found in 6 (r2 D0.4, P D 0.002). In
more acidic solutions, the dyes precipitate, and a red-shift of the absorbance

284
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
Fig. 5. Dependency of two selected dyes, 2d and 6 of the pH value of the aqueous environment. Shown is
the mean and SD of three independent measurements.
maximum takes place. In basic solutions of pH 10 and above, some quenching takes
place in 2d, while 6 remains pH-insensitive up to pH 12 (data not shown).
4. Discussion
Molecular rotors are a group of Xuorescent molecules that are characterized by a
dual deexcitation pathway. Molecular rotors in the excited singlet state can return to
the ground state either through Xuorescence emission or though the rotation of one
molecule's subunit against the other around a single bond. Due to the ability to
perform intramolecular rotation those molecules are also known as twisted intra-
molecular charge transfer (TICT) complexes. Since Xuorescence emission and intra-
molecular rotation are competitive processes, the Xuorescence quantum yield ꢃ
depends strongly on the intramolecular rotation rate (Eq. (1)),
where kr is the radiative relaxation rate and knr the non-radiative (rotational) relaxa-
tion rate. When Wrst discovered, TICT molecules such as 4-dimethylaminobenzonit-
rile (DMABN) were of high relevance because the intramolecular rotation rate
depends on the polarity of the solvent [30]. Some molecular rotors are less polarity
sensitive; primary determinant of intramolecular rotation and thus the quantum
yield is the solvent's viscosity. The relationship has Wrst been described by Förster
and HoVmann [23] following Eq. (2),
where ꢃ is the solvent viscosity, x is a dye-dependent constant, and C is a tempera-
ture-dependent constant. While changes in the solvent's polarity causes a change in
the Stokes shift, the emission intensity depends on the solvent viscosity. Those mole-
cules have received increasing attention as microviscosity sensors in polymerization

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
285
processes [26], viscosity indicators in phospholipid bilayers and the cell membrane
[24,31] and to measure bioXuid viscosity in a non-mechanical manner [32]. The last
application is of particular relevance, because traditional mechanical measurement of
Xuid viscosity is a time-consuming process, suVers from protein deposition at the
instrument walls and from artefacts caused by protein crosslinking at the Xuid–air
interface. Recent studies indicate that Xuorescence-based measurement of Xuid vis-
cosity with the precision of mechanical instruments is feasible. In addition, micro-
scopic measurement of a rotor's Xuorescence emission allows microviscosity
measurement with high spatial resolution. Readily available molecular rotors, with
the exception of CCVJ, are insoluble in water. For the determination of viscosity or
microviscosity in aqueous environments it is therefore desirable to endue rotor deriv-
atives that are readily soluble in water and aqueous liquids. In this study, we pre-
sented several new Xuorescent molecules with rotor characteristics that are soluble in
water and aqueous solutions.
Due to the constant C in Eq. (2), which reXects complex external inXuences, abso-
lute viscosity measurement is not directly possible. Relative viscosity changes, for
example the ratio of viscosities before and after a stimulus (ꢃ1 and ꢃ2, respectively),
can be computed quantitatively from the corresponding peak intensities I1 and I2
using Eq. (3) under the assumption that C does not change during the experiment [4]:
The constituents of complex Xuids, for example, dextran or proteins, also inXuence
the constants in Eq. (2). It is possible to calibrate the Xuorescence-based viscosity
measurement using similar Xuids with known viscosity [29,32], but individual calibra-
tions need to be performed for diVerent compositions. For example, diVerent calibra-
tion curves were found in aqueous solutions of dextran and hydroxyethylstarch
probed with 6 [29]. Depending on the Xuid, chemical interaction with the constituents
cannot be fully excluded.
All new compounds exhibited spectral properties closely related to 1 and 6.
Absorption took place in the blue range between 420 and 460nm, and light is emitted
in the green range, generally between 480 and 510nm. The Stokes shift was typically
40nm. A trend can be seen that julolidine-based dyes (primarily 2c and 2d) are
closely related to 1, while single-ring structures (10, 11) showed a blue-shift of the
excitation peak over 1, and a less pronounced shift of the emission peak towards
the blue. The single-ring structure shows some similarity to the well-explored 4-(N,
N-dimethylamino)-benzonitrile (DMABN), which is excited at 267 nm in the UV
range and which emits blue light. While DMABN is primarily known for its TICT-
induced second Xuorescence band in polar solvents [30], the compounds described in
this paper are primarily aimed at quantum yield changes with the viscosity of the
environment. In environments of diVerent polarity, only minor spectral shifts of less
than 5nm were seen (data not shown). Minor excitation blue-shifts were also noted
at high dye concentrations above 25 ꢂM. We attribute this to dye agglomeration,
which shields some of the dye from the polar environment. This eVect deWnes an
upper concentration limit for the use of these dyes. Dye 10 exhibits a relatively large

286
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
plateau in the excitation spectrum. Most likely this is caused by a double absorption
maximum, similar to the ones found in related dyes [28], but more closely spaced so
that the peaks are indistinguishable.
All dyes exhibited increased emission intensity in viscous Xuids as can be expected
from viscosity-sensitive molecular rotors. In a standard experiment [33], mixtures of
ethylene glycol and glycerol were used to modulate viscosity over a wide range (from
14mPas for pure ethylene glycol up to approximately 800mPas for pure glycerol)
without signiWcantly changing the solvent's polarity. As shown in Table 2 and Fig. 3,
the dyes presented in this study exhibited similar behavior, a power-law relationship
between intensity and viscosity that is well described by Eq. (2). Two empirical
constants exist in Eq. (2), the dye-dependent constant x [23] and the constant C that
is temperature-dependent. In our experiments we assumed a proportional relation-
ship of the measured peak emission intensity I with the quantum yield ꢃ, where the
proportionality is determined by such factors as excitation light intensity, geometry,
and detector eYciency. These external inXuences can be factored into a modiWed con-
stant C, allowing to determine x from the slope of intensity over viscosity in double-
logarithmic scale. This constant x can be interpreted as the dye's sensitivity towards
viscosity, because higher values of x amplify the intensity increase with viscosity.
Dyes 2c, 2d, and 10 exhibited the highest values for x, all of them very close to the
theoretical value of 0.6 expected for DCVJ (1) [23]. The primary cause for lower val-
ues of x in dyes 2b and 11 was a saturation eVect at high viscosities, where intensity
no longer followed the power-law relationship at the highest viscosity. This eVect led
to both reduced slope x and relatively poor regression data in the double-logarithmic
Wt, and potentially puts an upper limit to the viscosity that can be sensed to around
500mPa s. For comparison, full blood viscosity is approximately 5.7mPas, ethylene
glycol 14mPa s, and glycerol above 850 mPa s.
DiVerent mechanisms were employed to make the dyes hydrophilic. While 2c and
2d build on ethylene glycol to achieve hydrophilic properties, 10 presents hydroxyl
groups to the environment, and 2b features a COOH group much like 6. In the case
of dye 11, which does not have polar groups, the substitution of the tricyclic structure
for a single ring and an alkyl group was suYcient to ensure solubility. This eVect was
not observed in a comparable julolidine-based methyl ester (2e), which was tried else-
where [28] and found insoluble in water.
Since this study was aimed at the synthesis of molecular rotors that may act as vis-
cosity sensors in aqueous solutions, viscosity sensitivity was veriWed in water–colloid
systems similar to 6 as described in [29]. Dextran is a large starch molecule that can
be used in diVerent concentrations to increase the viscosity of aqueous solutions.
Dyes with high water solubility, 2d, 10, and 11, exhibit power-law relationship
between peak emission intensity and dextran-modulated viscosity. Compared to mix-
tures of ethylene glycol and glycerol, however, the slope x was markedly lower. This
is consistent with Wndings using 6 [29], where dextran solutions exhibited the lowest
slope among diVerent starches. We hypothesize that this behavior may be caused by
direct non-covalent interaction of the dye with hydrophobic sections of the starch
molecules, an eVect that would withdraw some of the dye from the free-volume inter-
action required for the viscosity-sensing process.

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
287
Intensity was compared with mechanical viscosity data. These measurements are
diVerent primarily because the Xuid is sheared to obtain its viscosity in the rheome-
ter, while no shear is applied when measuring Xuorescence. This needs to be taken
into account, because some Xuids (particularly solutions of macromolecules) may
exhibit non-Newtonian behavior. In our experiments, mixtures of ethylene glycol
and glycerol were considered Newtonian, while dextran solutions showed some
shear-thinning. Viscosity values were taken at a point where an increase of shear
rate showed no further reduction in the apparent viscosity. Thus, Xuorescence emis-
sion intensity was correlated to asymptotic viscosity. The fundamental relationship
between intensity and viscosity, however, would not be changed for other viscosity
points, e.g., extrapolated values at zero shear rate. Only the constant C in Eq. (2)
would change, corresponding to a parallel shift of the regression line in the double-
logarithmic plot.
Compound 10 and its precursors, 8 and 9, are chiral. In this study, they have been
synthesized as racemic mixture but the presented synthesis can be used to obtain
enantiomerically pure forms. Since most biological environments, for example, dex-
tran solutions, the cytosol, cell membranes, are chiral, it is conceivable that the two
antipodes of 10 could oVer additional information about a possible enantiospeciWc
nature of viscosity in biological media.
An important result that has not yet been extensively reported is the dye behavior
in solutions of diVerent pH. If molecular rotors are to be used as viscosity sensors in
bioXuids, the inXuence of proton concentration on the emission intensity in the phys-
iological range should be known. Our data indicate that in the range from pH 4 to
pH 10, only small intensity variations can be observed. Deviations remain within +7/
¡11% of the average intensity. No statistically signiWcant trend can be seen in 2d. In
even more acidic solutions, however, several eVects can be observed. A spectral red-
shift and reduced intensity (6) indicates energy loss by solvent interaction as well as
quenching by protonation. Solubility is also reduced, preventing solutions of concen-
trations higher than 1ꢂM. It can be concluded, however, that in the physiological
range centered on pH 7.5, pH inXuences do not play a signiWcant role.
5. Conclusion
In this paper we introduced several new molecular rotors aimed at aqueous solu-
tions, and demonstrated how twisted intramolecular charge transfer complex mole-
cules can be designed from a julolidine group and diVerent polar groups that ensure
solubility. Of the several compounds tested, 2d and 10 were of particular interest with
a combination of high viscosity sensitivity (high exponent x in Eq. (2)), high solubil-
ity in water, and demonstrated sensitivity to viscosity in aqueous colloid solutions.
Up to this point, most molecular rotors, namely the well-known DCVJ (1), were
insoluble in water. The fundamental diVerence between 6 and the new dyes 2d and 10
is the absence of a carboxyl group. While 2d builds on ethylene glycol to achieve
hydrophilic properties, 10 presents hydroxyl groups to the environment. These diVer-
ences may be exploited to allow or disallow speciWc reactions with the solvent, such

288
M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
as ester formation. These dyes show promise for use as hydrophilic viscosity sensors
in bioXuids, such as blood plasma and mixtures of plasma and colloids [32]. Future
studies may involve a covalently bound Xuorescent non-rotor unit to serve as a
reference, leading to a ratioing dye that eliminates sample inXuences and facilitates
calibration.
Acknowledgments
We gratefully acknowledge Wnancial support through NIH Grant 1R21-
RR018399. We thank the reviewers for helpful comments, particularly the question
of calibrating the intensity measurements, and the note on the additional use of the
enantiomers of 10.
References
[1] W. Möller, S. Takenaka, M. Rust, W. Stahlhofen, J. Heyer, J. Aerosol. Med. 10 (1997) 171–186.
[2] T.P. Stossel, Science 260 (1993) 1086–1094.
[3] P.F. Davies, Physiol. Rev. 75 (1995) 519–560.
[4] M.A. Haidekker, N. L'Heureux, J.A. Frangos, Am. J. Physiol. Heart Circ. Physiol. 278 (2000) H1401–
H1406.
[5] D.B. Goldstein, Annu. Rev. Pharmacol. Toxicol. 24 (1984) 43–64.
[6] R.A. Harris, R.J. Hitzemann, Curr. Alcohol 8 (1981) 379–404.
[7] R.A. Harris, P. Bruno, J. Neurochem. 44 (1985) 1274–1281.
[8] D.A. Johnson, R. Cooke, H.H. Loh, Adv. Exp. Med. Biol. 126 (1980) 65–68.
[9] G. Deliconstantinos, V. Villiotou, J.C. Stavrides, Biochem. Pharmacol. 49 (1995) 1589–1600.
[10] M.M. Gleason, M.S. Medow, T.N. Tulenko, Circ. Res. 69 (1991) 216–227.
[11] O. Nadiv, M. Shinitzky, H. Manu, D. Hecht, C.T. Roberts Jr., D. LeRoith, Y. Zick, Biochem. J. 298
(Pt. 2) (1994) 443–450.
[12] W. Osterode, C. Holler, F. Ulberth, Diabet. Med. 13 (1996) 1044–1050.
[13] M. Shinitzky, Biochim. Biophys. Acta 738 (1984) 251–261.
[14] J. Harkness, Biorheology 8 (1971) 171–193.
[15] R.L. Letcher, S. Chien, T.G. Pickering, J.E. Sealey, J.H. Laragh, Am. J. Med. 70 (1981) 1195–1202.
[16] M.A. McGrath, R. Penny, J. Clin. Invest. 58 (1976) 1155–1162.
[17] D.O. Bates, J.R. Levick, P.S. Mortimer, Clin. Sci. (Lond.) 85 (1993) 737–746.
[18] H. Schad, H. Brechtelsbauer, Z. Lymphol. 10 (1986) 14–19.
[19] M. Intaglietta, Artif. Cells Blood Substit. Immobil. Biotechnol. 22 (1994) 137–144.
[20] H.H. Lipowsky, J.C. Firrell, Am. J. Physiol. 250 (1986) H908–H922.
[21] A.G. Tsai, M. Intaglietta, Biorheology 38 (2001) 229–237.
[22] G.R. Cokelet, in: Y.C. Fung, N. Perrone, M. Anliker (Eds.), Biomechanics—Its Foundations and
Objectives, Prentice-Hall, Englewood CliVs, 1972, pp. 63–103.
[23] Th. Förster, G. HoVmann, Z. Phys. Chem. 75 (1971) 63–76.
[24] C.E. Kung, J.K. Reed, Biochemistry 25 (1986) 6114–6121.
[25] R.O. Loutfy, B.A. Arnold, J. Phys. Chem. 86 (1982) 4205–4211.
[26] R.O. Loutfy, Pure Appl. Chem. 58 (1986) 1239–1248.
[27] M. Haidekker, T. Brady, K. Wen, C. Okada, H. Stevens, J. Snell, J. Frangos, E. Theodorakis, Bioorg.
Med. Chem. 10 (2002) 3627–3636.
[28] M.A. Haidekker, T. Ling, M. Anglo, H.Y. Stevens, J.A. Frangos, E.A. Theodorakis, Chem. Biol. 8
(2001) 123–131.
[29] W. Akers, M.A. Haidekker, J. Biomech. Eng. 126 (2004) in press.

M.A. Haidekker et al. / Bioorganic Chemistry 32 (2004) 274–289
289
[30] Z.R. Grabowski, J. Dobkowski, Pure & Appl. Chem. 55 (1983) 245–252.
[31] M.L. Viriot, M.C. Carré, C. GeoVroy-Chapotot, A. Brembilla, S. Muller, J.-F. Stoltz, Clin. Hemorheol.
Microcirc. 19 (1998) 151–160.
[32] M.A. Haidekker, A.G. Tsai, T. Brady, H.Y. Stevens, J.A. Frangos, E. Theodorakis, M. Intaglietta, Am.
J. Physiol. Heart Circ. Physiol. 282 (2002) H1609–H1614.
[33] T. Iwaki, C. Torigoe, M. Noji, M. Nakanishi, Biochemistry 32 (1993) 7589–7592.