Polarity-sensitive coumarins tailored to live cell imaging

Article abstract of DOI:10.1021/ja9050444

Polarity-dependent fluorescent probes are recently attracting interest for high-resolution cell imaging. Following a stepwise rational approach, we prepared and tested a toolbox of new coumarin derivatives tailored to in vivo imaging applications. Our com

Full text of DOI:10.1021/ja9050444

Published on Web 01/05/2010
Polarity-Sensitive Coumarins Tailored to Live Cell Imaging
Giovanni Signore,*,† Riccardo Nifos`ı,‡ Lorenzo Albertazzi,†,‡ Barbara Storti,†,‡ and
Ranieri Bizzarri*,†,‡
NEST, CNR-INFM, p.za San SilVestro 12, 56127 Pisa, Italy, NEST, Scuola Normale Superiore, and
IIT@NEST, Center for Nanotechnology InnoVation, p.za San SilVestro 12, I-56127 Pisa, Italy
Received June 19, 2009; E-mail: giovanni.signore@iit.it; r.bizzarri@sns.it
Abstract: Polarity-dependent fluorescent probes are recently attracting interest for high-resolution cell
imaging. Following a stepwise rational approach, we prepared and tested a toolbox of new coumarin
derivatives tailored to in vivo imaging applications. Our compounds are characterized by a donor-(coumarin
core)-acceptor molecular structure, where the electron donor is represented by alkylether or naphthyl
groups, and the electron acceptor is represented by benzothiazene and cyano groups. Prior to synthesis,
the substitution patterns were screened by computational methods to provide functional fluorescent
derivatives easy to synthesize, and with excitation in the visible region of spectrum. We set up a robust
synthetic procedure tunable on the substitution patterns to achieve. These coumarins possess excellent
fluorescence quantum yields (up to 0.95), high molar extinction coefficients (up to 46,000 M-1 cm-1), and
large Stokes shifts. Furthermore, they display strong solvatochromism, being almost non-emissive in water
and very fluorescent in less polar media (up to 780-fold enhancement in brightness). The solvatochromism
of these compounds can be accounted for by a photophysical method encompassing two communicating
excited states. When tested on cultured cells, the results showed that the developed coumarins were not
harmful and their photophysical properties were unchanged compared to free solution. According to the
determined solvatochromic properties, the coumarin fluorescence was detected only in the most lipophilic
environments of the cell. The prepared compounds represent remarkable tools to investigate subtle
biochemical processes in the cell environment after appropriate conjugation to biomolecules, and at the
same time constitute the basis for further engineering of a new generation of biosensors.
1. Introduction
molecular cargo and/or to bind to a specific protein configuration
are now widely investigated for this purpose.5 How to signal
the protein recognition event, however, is still unclear. An
elegant answer to this question considers organic dyes whose
optical properties are sensitiVe to the surrounding environment
(solvatochromic probes).6-19 According to this approach, the
Cell behavior is regulated by transient activation of protein
activities in specific subcellular regions. The recent development
of highly fluorescent probes tailored to the cellular environment
and of high-resolution imaging techniques allows monitoring
protein activity in living cells by direct optical methods.1
Members of the green fluorescent protein family (GFPs) are
particularly impressive nanoprobes for in ViVo studies, owing
to their genetically encoded fluorescence.2,3 Effective chemical
methods were developed to link functional organic dyes, often
more fluorescent and resistant to photodegradation than GFPs,
to proteins in ViVo.4 In most cases, however, genetic modification
of the target protein is needed for the probe binding; the least
invasive methods involve adding non-native peptide sequences
of 6-20 amino acids to the target protein. Thus, the detection
of protein activity in the true native state or after post-
translational modification is intrinsically impossible. The use
of multicomponent molecular species penetrating into the cell
and capable of recognizing selectively a certain state of the target
protein (universal biosensors) could overcome this problem.
Peptide sequences able to convey into the cell a tailored
(5) Cardarelli, F.; Serresi, M.; Bizzarri, R.; Beltram, F. Traffic 2008, 9,
528–539.
(6) Toutchkine, A.; Kraynov, V.; Hahn, K. M. J. Am. Chem. Soc. 2003,
125, 4132–4145.
(7) Nalbant, P.; Hodgson, L.; Kraynov, V.; Toutchkine, A.; Hahn, K. M.
Science 2004, 305, 1615–1619.
(8) Cohen, B. E.; Pralle, A.; Yao, X.; Swaminath, G.; Gandhi, C. S.; Jan,
Y. N.; Kobilka, B. K.; Isacoff, E. Y.; Jan, L. Y. Proc. Natl. Acad.
Sci. U.S.A. 2005, 102, 965–970.
(9) Kobiilka, B. K.; Gether, U. Methods Enzymol. 2002, 343, 170–182.
(10) Mannuzzu, L. M.; Moronne, M. M.; Isacoff, E. Y. Science (Wash-
ington, DC, U.S.) 1996, 271, 213–216.
(11) Hanouni, P.; Gryczynski, Z.; Steenhuis, J.; Lee, T. W.; Farrens, D. L.;
Lakowicz, J. R.; Kobilka, B. K. J. Biol. Chem. 2001, 276, 24433–
24436.
(12) Cohen, B. E.; McAnaney, T. B.; Park, E. S.; Jan, Y. N.; Boxer, S. G.;
Jan, L. Y. Science 2002, 296, 1700–1703.
(13) Lavis, L. D.; Raines, R. T. ACS Chem. Biol. 2008, 3, 142–155.
(14) Venkatraman, P.; Nguyen, T. T.; Sainlos, M.; Bilsel, O.; Chitta, S.;
Imperiali, B.; Stern, L. Nat. Chem. Biol. 2007, 3, 222–228.
(15) Fu¨lo¨p, A.; Arian, D.; Lysenko, A.; Mokhir, A. Bioorg. Med. Chem.
Lett. 2009, 19, 3104–3107.
† IIT@NEST, Center for Nanotechnology Innovation, I-56127 Pisa, Italy.
‡ NEST, Scuola Normale Superiore and CNR-INFM, I-56127 Pisa, Italy.
(1) Petty, H. R. Microsc. Res. Tech. 2007, 70, 687–709.
(2) Miyawaki, A. Neuron 2005, 48, 189–199.
(16) Sainlos, M.; Iskenderian, W. S.; Imperiali, B. J. Am. Chem. Soc. 2009,
131, 6680–6682.
(3) Bizzarri, R.; Arcangeli, C.; Arosio, D.; Ricci, F.; Faraci, P.; Cardarelli,
F.; Beltram, F. Biophys. J. 2006, 90, 3300–3314.
(17) Dai, Z.; Dulyaninova, N. G.; Kumar, S.; Bresnick, A. R.; Lawrence,
D. S. Chem. Biol. 2007, 1254–1260.
(4) Prescher, J. A.; Bertozzi, C. R. Nat. Chem. Biol. 2005, 1, 13–21.
9
1276 J. AM. CHEM. SOC. 2010, 132, 1276–1288
10.1021/ja9050444  2010 American Chemical Society

Live Cell Imaging with Solvatochromic Coumarins
A R T I C L E S
recognition signal comes from a solvatochromic prosthetic group
able to sense the polarity change associated with binding.
Reasonably, it should be located at a site close enough to the
binding region of the biosensor (where changes in polarity
necessarily occur) but not too close, so that it does not hamper
the recognition and binding process.
In past years, solvatochromic dyes have been extensively
employed in selective staining of many subcellular domains.
Notably, solvatochromic FM dyes are widely employed as
membrane-staining agents, since they localize preferentially on
cell membrane, where their fluorescence is enhanced.20 How-
ever, the development of solvatochromic dyes as polarity probes
to investigate protein binding, membrane rearrangement, and
other biological processes has not evolved to its full potential.
Fluorescent coumarins are attractive fluorophores, as they are
characterized by high quantum yields (up to 0.90),21 high
extinction coefficients (10,000-40,000),22,23 large Stokes shifts
(up to 160 nm),24 and can be engineered to respond to their
environment polarity (solvatochromic probes).25-27 These prop-
erties promoted their use as dyes,28,21,29 sensors for metal
cations,30-34 and dopants for OLEDs.35,36 The well-known
commercial fluorophores Alexa 350 and Alexa 430 are based
on a coumarin core.37 Recently, coumarins were thoroughly
investigated in view of synthesizing new probes for biological
applications, in particular for the imaging of living cells.38-48
In spite of their relevance as biosensors, however, no studies
describing solvatochromic coumarins tailored to probe the
different polarity properties of biological environments in
cultured cells have been yet reported.
A powerful method to confer solvatochromic properties to
aromatic fluorophores involves the concomitant functionalization
with electron-donor (D) and electron-acceptor (A) groups. This
effect is related to the large dipole moment present in such
structures and its significant variations between the ground and
the excited state. In such a case, the excited state is often referred
to as intramolecular charge transfer (ICT). The change in
solvation energy between polar and apolar environments ac-
counts for the shift in intensity and/or wavelength in their
absorption and fluorescence spectra.49 Alternatively, the ICT
state can evolve along the energy landscape to excited states of
different nature following a process that is strongly influenced
by the surrounding polarity.
The aromatic core of coumarin is particularly suitable for
the introduction of D and A groups, as the most common
synthetic pathway involves the use of hydroxy aldehydes and
2-aryl substituted acetates. Both of these aromatic subunits can
be easily functionalized with A or D groups. This work will
present the synthesis and spectroscopic analysis of new solva-
tochromic coumarins for intracellular use bearing one A group
in position 3, and D groups in positions 5, 6, 7 and/or 8.
Concerning position 3, our choice fell on a benzothiazene
group, owing to its electron-withdrawing aromatic system
capable also to extend electron conjugation. Amino substituents
are excellent D groups, and would represent a good choice for
the electron-rich part of the fluorophore. These substituents,
however, bring about significant drawbacks. Indeed, aromatic
amino groups are quite reactive and may undergo chemical
processes that hamper their further functionalization and/or
strongly affect the solvatochromic outcome (e.g., protonation
of the amino group strongly decreases its electron-donating
capability). Our selection for the positions 5-8 fell on oxygen-
based D groups, which essentially overcome these limitations
while retaining most of the electron-donating properties of amino
substituents. Additionally, we explored the possibility to
incorporate a further aromatic ring to increase electronic
conjugation. In the following, we shall show that these new
fluorophores display high sensitivity to the environmental
polarity combined with good-to-excellent brightness and are
indeed suitable for intracellular use.
(18) Loving, G.; Imperiali, B. J. Am. Chem. Soc. 2008, 130, 13630–13638.
(19) Volodymyr, S. V.; Klymchenko, A. S.; de Rocquigny, H.; Me´ly, Y.
Nucleic Acids Res. 2009, 37, e25.
(20) Bolte, S.; Talbot, C.; Boutte, Y.; Catrice, O.; Read, N. D.; Satiat-
Jeunemaitre, B. J. Microsc. 2004, 214, 159–173.
(21) Turki, H.; Abid, S.; Fery-Forgues, S.; El Gharbi, R. Dyes Pigm. 2007,
73, 311–316.
(22) Christie, R. M.; Lui, C.-H. Dyes Pigm. 2000, 47, 79–89.
(23) Wolfbeis, O. S.; Koller, E.; Hochmuth, P. Bull. Chem. Soc. Jpn. 1985,
58, 731–734.
(24) Elangovan, A.; Lin, J.-H.; Yang, S.-W.; Hsu, H.-Y.; Ho, T.-I. J. Org.
Chem. 2004, 69, 8086–8092.
(25) Hirano, T.; Hiromoto, K.; Kagechika, H. Org. Lett. 2007, 9, 1315–
1318.
(26) Turki, H.; Abid, S.; El Gharbi, R.; Fery-Forgues, S. C. R. Chimie
2006, 9, 1252–1259.
(27) Uchiyama, S.; Takehira, K.; Yoshihara, T.; Tobita, S.; Ohwada, T.
Org. Lett. 2006, 8, 5869–5872.
(28) Christie, R. M.; Morgan, K. M.; Islam, M. S. Dyes Pigm. 2008, 76,
741–747.
(29) Luo, X.; Naiyun, X.; Cheng, L.; Huang, D. Dyes Pigm. 2001, 51,
153–159.
2. Results
(30) Peng, M.-S.; Cai, J. Dyes Pigm. 2008, 79, 270–272.
(31) Wang, J.; Qian, X.; Cui, J. J. Org. Chem. 2006, 71, 4308–4311.
(32) Lin, W.; Yuan, L.; Feng, J.; Cao, X. Eur. J. Org. Chem. 2008, 73,
2689–2692.
2.1. Computational Screening of Coumarin Compounds.
Preliminarily to any experimental step, a computational analysis
by density functional theory (DFT/B3LYP) was performed to
identify interesting substitution donor-(coumarin core)-acceptor
(33) Lim, N. C.; Schuster, J. V.; Porto, M. C.; Tanudra, M. A.; Yao, L.;
Freake, H. C.; Bruckner, C. Inorg. Chem. 2005, 44, 2018–2030.
(34) Ma, Y.; Luo, W.; Quinn, P. J.; Liu, Z.; Hider, R. C. J. Med. Chem.
2004, 47, 6349–6362.
(35) Lee, M.-T.; Yen, C.-K.; Yang, W.-P.; Chen, H.-H.; Liao, C.-H.; Tsai,
C.-H.; Chen, C. H. Org. Lett. 2004, 6, 1241–1244.
(36) Swanson, S. A.; Wallraff, G. M.; Chen, J. P.; Zhang, W.; Bozano,
L. D.; Carter, K. R.; Salem, J. R.; Villa, R.; Campbell, S. Chem. Mater.
2003, 15, 2305–2312.
(42) Kim, H. M.; Fang, X. Z.; Yang, P. R.; Yi, J.-S.; Ko, Y. G.; Piao,
M. J.; Chung, Y. D.; Park, Y. W.; Jeon, S.-J.; Cho, B. R. Tetrahedron
Lett. 2007, 48, 2791–2795.
(43) Dillingham, M. S.; Tibbles, K. L.; Hunter, J. L.; Bell, J. C.;
Kowalczykowski, S. C.; Webb, M. R. Biophys. J. 2008, 95, 3330–
3339.
(37) Panchuk-Voloshina, N.; Haugland, R. P.; Bishop-Stewart, J.; Bhalgat,
M. K.; Millard, P. J.; Mao, F.; Leung, W.-L.; Haugland, R. P.
J. Histochem. Cytochem. 1999, 47, 1179–1188.
(44) Webb, M. R.; Corrie, J. E. T. Biophys. J. 2001, 81, 1562–1569.
(45) Katritzky, A. R.; Cusido, J.; Narindoshvili, T. Bioconjugate Chem.
2008, 19, 1471–1475.
(46) Komatsu, K.; Urano, Y.; Kojima, H.; Nagano, T. J. Am. Chem. Soc.
2007, 129, 13447–13454.
(47) Komatsu, H.; Miki, T.; Citterio, D.; Kubota, T.; Shindo, Y.; Kitamura,
Y.; Oka, K.; Suzuki, K. J. Am. Chem. Soc. 2005, 127, 10798–10799.
(48) Webb, W. R.; Corrie, J. E. T. Biophys. J. 2001, 81, 1562–1569.
(49) Suppan, P.; Ghoneim, N. SolVatochromism; The Royal Society of
Chemistry, Cambridge, U.K., 1997; ISBN 0-85404-419-1.
(38) Brun, M.-P.; Bischoff, L.; Garbay, C. Angew. Chem., Int. Ed. 2004,
43, 3432–3436.
(39) Zhao, Y.; Zheng, Q.; Dakin, K.; Xu, K.; Martinez, M. L.; Li, W.-H.
J. Am. Chem. Soc. 2004, 126, 4653–4663.
(40) Eckardt, T.; Hagen, V.; Schade, B.; Schmidt, R.; Schweitzer, C.;
Bendig, J. J. Org. Chem. 2002, 67, 703–710.
(41) Coleman, R. S.; Berg, M. A.; Murphy, C. J. Tetrahedron 2007, 63,
3450–3456.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1277

A R T I C L E S
Signore et al.
Table 1. Calculated and Measured Maximum Absorption Wavelengths of 1a-g
a Planar geometry, calculated by time-dependent DFT (B3LYP/6-31G*>), PCM acetonitrile. Values in parentheses indicate the oscillator strength;
b Twisted geometry, optimized by Hartree-Fock method, optical values calculated by time-dependent DFT (B3LYP/6-31G*). Values in parentheses
indicate the oscillator strength. c Experimental, solvent: acetonitrile.
patterns onto the basic coumarin rings as well as to put in
evidence the differences in the spectroscopic properties of these
structures. The performance of DFT methods on coumarins has
been already tested.50,51 For structure optimization we also
compared the DFT results with those obtained by the Hartree-
Fock method, because DFT has been reported to artificially
predict planar structures in molecules where the rotational energy
barriers are low.52 All calculations were performed with the
Gaussian 03 package.53
The whole set of screened structures is reported in Scheme
1 in Supporting Information.
Our attention was eventually attracted by seven different
compounds carrying dialkyl or naphthyl groups in position 6-8,
benzothiazene in position 3, and cyano groups in position 4 of
the coumarin ring. Indeed, the benzothiazenyl group was
predicted to generate a large red-shift of the excitation wave-
length, owing to both the extension of the electronic conjugation
and the electron-withdrawing effect of the added aromatic ring.
Additionally, disubstituted coumarins seem very promising in
terms of further functionalization to yield activated “probes”
for conjugation to proteins and other biomolecules.
(50) Preat, J.; Jacquemin, D.; Perpete, E. A. Chem. Phys. Lett. 2005, 415,
20–24.
(51) Xu, B.; Yang, J.; Jiang, X.; Wang, Y.; Sun, H.; Yin, J. J. Mol. Struct.
2009, 917, 15–20.
Both DFT/B3LYP and Hartree-Fock revealed that all non-
cyano compounds share a planar conformation where the sulfur
of benzothiazene and the ring carboxyl group of coumarin are
arranged on the same side (the other conformation, where they
(52) Zhao, K.; Ferrighi, L.; Frediani, L.; Wang, C.-K.; Luo, Y. J. Chem.
Phys. 2007, 126, 204509–204515.
(53) Frisch, M. J. et al. Gaussian 03, Revision B.03; Gaussian, Inc.:
Wallingford, CT, 2004.
9
1278 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Live Cell Imaging with Solvatochromic Coumarins
A R T I C L E S
Scheme 1. Synthesis of 3c
Scheme 2. Synthesis of Coumarins; Substituents: R1 ) R4 ) H, R2
) R3 ) OMe (a,e); R1 ) R2 ) H, R3 ) R4 ) OMe (b,f); R1 ) R2 )
H, R3,R4 ) -O-CH2-CH(COOEt)-O-(c,g); R1,R2 ) fused phenyl
ring, R3 ) R4 ) H (d); R1 ) R2 ) H, R3 ) R4 ) OH (4)
are opposite, is 4-5 kcal/mol higher in energy). In contrast,
Hartree-Fock, but not B3LYP, predicted a twisted geometry
with an angle of 20-30° between the planes of the coumarin
and the benzothiazene rings for the cyano compounds.
The excitation wavelengths and the oscillator strengths
calculated by time-dependent DFT (B3LYP/6-31G*) for the
planar and the twisted geometries (the latter case only for cyano
coumarins) are reported in Table 1. The values are calculated
including implicit solvent effects via the polarizable continuum
field method (PCM)54 using acetonitrile as reference solvent
with high dielectric constant. The DFT analysis shows a quite
strong transition, mainly of π-π* character, for all compounds.
Excitation wavelengths range in the 390-420 nm and 460-470
nm regions for compounds without and with cyano groups,
respectively. Hence, the presence of electron-donating/with-
drawing groups red-shifts the π-π* transition by 70-180 nm
compared to the basic coumarin ring (310 nm). For 1g DFT
predicts additional peaks in the >400 nm region due to excitation
to higher excited states.
2.3. General Optical Properties of the Coumarin Compounds.
Absorption and fluorescence spectra in acetonitrile contain all
the principal optical features of the prepared coumarins found
also in other solvents and need to be described first. Furthermore,
acetonitrile data allow for a comparison with spectroscopic
characteristics determined by DFT. The absorption and excita-
tion/emission spectra of representative compounds 1a, 1c, 1d
and 1e are shown in Figure 1, whereas quantitative data can be
found in Table 2.
The absorption maxima of coumarins devoid of the cyano
group were found to be in excellent agreement with the
theoretical values (Table 1). These compounds showed large
absorption bands (in some cases also retaining vibrational
substructure) peaking in the 380-410 nm interval (Figure
1a-c). Furthermore, these compounds were found to possess
broad fluorescence emission spectra spanning 400-600 nm
(Figure 1), associated with remarkably high fluorescence
quantum yields (Table 2). Excitation spectra were superimpos-
able on absorption spectra, indicating that the quantum yield is
independent of the excitation wavelength (data not shown).
Comparison of 1a and 1d (Figure 1a,c) suggests that two
methoxy groups (1a) or a further aromatic ring (1d) have
approximately the same electronic effect on both the funda-
mental and excited states on account of a substantial constancy
of absorption and emission wavelengths. Instead, the transfer of
a methoxy group from position 6 to 8 on the coumarin ring leads
to 10-20 nm blue-shifts of absorption and emission maxima (Table
2, compare 1a and 1b). Substitution of two methoxy groups with
one bidentate alkyl ether amenable to further functionalization
affects neither the absorption nor the emission properties of the
molecule (Table 2, compare 1b and 1c).
Apparently, the substitution regioselectivity plays a minor role
in determining the transition wavelength. A small red-shift is
also associated with the addition of a further aromatic ring
compared to alkylether/hydroxyl groups (Table 1, 1a and 1d).
Notably, the twisted geometry leads to higher transition energies
(i.e., shorter wavelengths). This trend was verified also for larger
angles (50-90°) between the two aromatic rings, although these
were predicted to be higher in energy (∼3 kcal/mol).
2.2. Synthesis of Coumarin Compounds. The first step in our
synthetic strategy concerned the synthesis of ethyl 2-(benzo-
[d]thiazol-2-yl)-acetate 2 starting from the o-aminothiophenol
and ethyl cyanoacetate. The reaction was carried out under
solvent-free conditions at 120 °C to ensure complete conversion
to the desired product.55 2 played the role of the first reactant
in the piperidine-catalyzed Knoevenagel condensation that we
designed and accomplished to obtain the final coumarins
(Scheme 2).56
The other reactants were substituted o-hydroxy aldehydes
bearing different functional groups (3a,b′,c,d). Notably, the two
noncommercial hydroxy aldehydes 3a and 3c were synthesized
through a Vilsmeier formylation reaction and an alkylation by
ethyl 2,3-dibromopropionate under basic conditions (Scheme
1), respectively. The Knoevenagel condensation was found to
proceed smoothly, and the products (1a,c,d, and 4) were
recovered in excellent yields by simple filtration of the reaction
mixture.
4 was further chemically modified in order to complete the
toolbox of coumarin chromophores. 1b was prepared by direct
methylation of 4 in nearly quantitative yield. Finally, a cyano
group was introduced in position 4 of coumarin 1a-c by an
addition/elimination reaction similar to that described in ref 23
(Scheme 2). The latter synthetic path led to the cyano coumarins
1e-g with yields around 80%. Coumarins 1a-g were analyti-
cally pure (>99%), as evidenced by HPLC analyses.
(54) Cossi, M.; Barone, V. J. Chem. Phys. 2001, 115, 4708–4718.
(55) Abbotto, A.; Bradamante, S.; Facchetti, A.; Pagani, G. A. J. Org.
Chem. 2002, 67, 5753–5772.
(56) Lee, S.; Sivakumar, K.; Shin, W.-S.; Xie, F.; Wang, Q. Bioorg. Med.
Chem. Lett. 2006, 16, 4596–4599.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1279

A R T I C L E S
Signore et al.
Table 2. Spectroscopic Data of Compounds 1a-g
H2O i-PrOH H2O/i-PrOH DMSO ACN OctOH AcOEt Diox H2O/Diox
1a
λmax 401.5 407
409.5
3.8
476
411
3.5
480
405
3.4
474
409.5 405.5 406
408.5
3.3
478
a
1.3
3.6
475
3.3
3.5
3.9
ε
λem 475
477
472
473
b
0.10 0.88
0.86
0.95 0.86 0.89 0.88 0.86
0.84
Φ
1b
λmax 381.0 389.5
393
3.6
470
0.65
393
3.7
471
386.5 392.5 387.5 387
389
4.3
470
0.55
a
0.4
3.7
465
3.7
3.6
4.5
4.1
ε
λem 487
464
466
460
461
b
0.01 0.74
0.78 0.86 0.75 0.79 0.78
Φ
1c
λmax 382
388.5
4.4
461
389.5
4.1
466
392
3.9
468
386.5 391.5 387
387
4.1
458
389
4.6
466
0.72
a
2.5
3.6
3.5
3.8
ε
λem 470
460
463
457
b
0.12 0.81
0.75
0.82 0.81 0.86 0.85 0.85
Φ
1d
λmax 403.5 411
414
1.9
482
0.92
413.5 406.5 413 407.5 408
412.5
2.7
475
a
1.0
2.8
478
2.8
2.8
3.1
3.3
3.3
ε
λem 481
486
476
479
474
484
b
0.04 0.89
0.86 0.87 0.87 0.83 0.83
0.86
Φ
1e
λmax 454.5 457.5
461
1.8
558
0.20
476
461.5 455.5 460 455.5 454.5 458.5
a
0.3
1.6
541
2.6
2.8
1.8
3.3
3.0
1.8
ε
λem 581
560
549
539
534
526
560
0.22
478
b
0.01 0.40
0.26 0.41 0.51 0.52 0.59
480
Φ
λsh 471
475
474
478
473
475
1f
λmax 438
428
2.7
521
433
2.2
nd
430
2.4
526
426
3.3
nd
433
2.4
529
424
2.2
546
424
2.8
526
428
2.1
nd
a
0.9
ε
λem nd
Φ
b
<0.01 0.01
<0.01
0.02 <0.01 0.02 0.01 0.02
<0.01
1g
λmax 430
427
2.2
518
431
1.5
529
0.02
427
1.4
520
424
2.0
nd
432
2.1
513
424
2.0
516
423
2.3
507
429
1.7
535
0.01
a
1.1
ε
λem 555
b
0.01 0.04
0.02 <0.01 0.1
0.09 0.14
Φ
a Molar extinction coefficient (×10-4). b Fluorescence quantum yield.
interval. Conversely, excitation of the high-energy absorption
shoulder (<370 nm) leads always to a composite emission
spectra encompassing two bands in nearly fixed intensity
ratio: the first band peaks at 470 nm (λsh), and the second is
identical to that obtained by excitation at absorption maxi-
mum. Importantly, this spectroscopic pattern, and particularly
the invariant intensity ratio between the two bands, was found
to be independent of fluorophore concentration, thus ruling
out the formation of excimers (Supporting Information). We
shall provide a more detailed explanation of these features
in the Discussion section.
Cyanocoumarins 1f,g display very poor quantum yields in
acetonitrile and in all other solvents (Table 2). Notably, DFT
calculations overestimated the actual absorption wavelength of
these cyanocoumarins (Tables 1 and 2). On account of the
calculated dependence of transition energy on the twist angle
between the two aromatic rings, this discrepancy may arise from
more distorted geometry than predicted by our theoretical
analysis. This condition could be at basis of the observed poor
emissivity, assuming that a distorted conformation promotes
nonradiative decay channels.
2.4. Solvatochromic Properties of the Coumarin Compounds.
Next, we set out to investigate the solvatochromic behavior of
compounds 1a-g. For this goal, we measured absorption and
fluorescence spectra at 25 °C in an ensemble of solvents
covering a large polarity interval as expressed by the orientation
Figure 1. Absorption (full line) and emission (dotted line, λexc ) 405 nm)
spectra in water (black), acetonitrile (red), and 1-octanol (blue) of coumarins
1a (a), 1c (b), 1d (c), 1e (d).
As expected, the introduction of a cyano group in position 4
of the ring leads to a significant absorption and emission red-
shift (Table 2, Figure 1d). 1e is characterized by easily detected
dual fluorescence emission (Figure 1d). Selective excitation
at the absorption maxima is associated with a unique broad
emission peak at 550 nm and covering the 480-700 nm
9
1280 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Live Cell Imaging with Solvatochromic Coumarins
A R T I C L E S
Figure 2. (a) Polarity dependence of emission quantum yield for 1a (red) and 1e (blue). (b) Relative brightness (absorption × quantum yield) enhancement
(Water ) 1) for coumarins 1a-e. (c) Brightness enhancement for 1a (bottom) and 1e (top) in water/i-PrOH (red) and water/Dioxane (blue) mixtures. (d)
Fluorescence emission of 1e upon excitation at 405 nm in different mixtures water/i-PrOH (v/v percentages of isopropanol are indicated in the legend). (e)
Fluorescence lifetimes of 1a (λexc: 403 nm) and 1e (λexc: 468 nm) in water, i-PrOH, and acetonitrile. (f) Polarity-dependence of fluorescence lifetimes of 1e
upon excitations at 403 nm (blue) and 468 nm (orange).
polarizability ∆f:57,58 water (symbol: W, ∆f ) 0.32), acetonitrile
(ACN, ∆f ) 0.306), isopropanol (i-PrOH, ∆f ) 0.273),
dimethylsulfoxide (DMSO, ∆f ) 0.265), ethyl acetate (EtOAc,
∆f ) 0.199), n-octanol (OctOH, ∆f ) 0.103), and 1,4-dioxane
(Diox, ∆f ) 0.021). 50/50 vol % mixtures of W/i-PrOH (∆f )
0.306) and W/Diox (∆f ) 0.295) were also added to provide
media with intermediate hydrogen-bonding capabilities. Very
apolar solvents such as n-hexane were avoided, as the coumarins
became rather insoluble. Spectroscopic data for the coumarins
in all solvents are reported in Table 2.
On the other hand, fluorescence emission of all compounds is
nearly abolished in water, as witnessed by the very low quantum
yields and lower extinction coefficients measured in this medium
(Table 2, Figure 1). Remarkably, quantum yields of non-
cyanocoumarins 1a-d do not correlate with the polarity function
(∆f), being rather constant in organic or water/organic media
and much lower in water (Figure 2a, red). A somewhat stronger
correlation between quantum yield and ∆f is observable for the
fluorescent cyanocoumarin 1e (Figure 2a, blue). All coumarins,
however, displayed no correlation at all between absorption (or
emission) wavelength and ∆f. As such a correlation would be
expected by the classical Lippert-Mataga’s theory of solvato-
chromism, these findings suggest a specific interaction between
water molecules and the coumarin structures that strongly
perturbs the photophysical behavior of the latter ones. The band
broadening of the absorption spectra in water is a further support
to this hypothesis.
On one hand, all coumarins display high extinction coef-
ficients and quantum yields in organic solvents or water/solvent
mixtures, resulting in bright fluorescence (Table 2, Figure 1).
(57) Suppan, P. J. Chem. Soc. A 1968, 3125–3133.
(58) Weast, R. C. Handbook of Chemistry and Physics; CRC Press: Boca
Raton, FL, 1980.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1281

A R T I C L E S
Signore et al.
Table 3. Fluorescence Lifetimes of 1a,c,e in Water, Isopropanol,
and Acetonitrile; Intrinsic Lifetime Values Calculated from the
Strickler-Berg Relation62 are Reported in Parentheses
The fluorophore brightness (i.e., the product of extinction
coefficient and quantum yield) is a parameter immediately
associated with the fluorescence emission performance in
environments of different polarity at constant excitation power,
as typically occurring in cell imaging. Accordingly, we com-
puted the absolute brightness (Supporting Information, Table
S1) and water-relative brightness (Figure 2b) of our coumarins.
Among non-cyanocoumarins, 1b displays an astonishing >350-
fold increment of brightness going from water to water/organic
or organic media. Nonetheless, 1a is associated with the highest
absolute brightness in non-water solvents (>28,000, Supporting
Information, Table S1). Cyanocoumarin 1e shows a quite
variable water-relative brightness (between 170-790), in keep-
ing with its more solvent-modulated extinction coefficient and
quantum yield.
fluorescence lifetime (ns)
coumarin H2O i-PrOH CAN
radiative lifetime (ns)
H2O
i-PrOH
ACN
1a
1c
1e
3.28 3.18 3.28
2.44 2.69 2.65
3.23 4.85 5.24 323.0 (36.2) 12.1 (8.28) 12.8 (5.29)
32.8 (6.62) 3.6 (3.13)
20.3 (4.03) 3.32 (2.52) 3.27 (3.03)
3.8 (3.19)
supports the use of 1e as a ratiometric (i.e., concentration
independent)61 sensor of medium polarity.
2.5. Fluorescence Lifetime Measurements. On account of
their non-negligible fluorescence in water, coumarin 1a,c were
selected to test the effect of solvent nature on the fluorescent
lifetime (τ). 1e was added to this pool to compare the
photophysical characteristics after the incorporation of the cyano
group. Actually, as already mentioned, 1e displays dim but
detectable fluorescence in water when excited at 405 nm.
Remarkably, 1a,c,e are all characterized by monoexponential
decays when excited near their absorption maximum (403 nm
for 1a,c, 468 nm for 1e) regardless of the solvent (Figure 2e).
Lifetime values range from 2.8 to 5.24 ns (Table 3). The actual
lifetimes can be combined with the quantum yields to give the
corresponding radiative lifetimes (τr, Table 3).
For 1a,c, the position and nature of the functional groups
attached to the coumarin ring modulate the detected lifetime.
Interestingly, for both coumarins τ is quite constant in the three
tested solvents, whereas τr becomes extremely large (>30 ns)
in water. The theoretical τr in water calculated by the
Strickler-Berg equation from the absorption and fluorescence
spectra is around 3 ns, an order of magnitude lower than the
experimental value. Instead, the theoretical τr in i-PrOH and
ACN are in good agreement with the experimental value (Table
3). The Strickler-Berg equation fails to yield reasonable values
of τr only when a strong interaction with the solvent takes place
and/or there is a change in the excited-state geometry. These
data suggest that the low quantum yields of 1a,c in water are
not the result of some quenching process but rather of different
properties of the excited state. A photophysical model account-
ing for this phenomenology will be described in the Discussion
section.
The cyano group in 1e leads to significantly longer τ and τr
(Table 3). Here, τr is significantly larger than the theoretical
value in all media, suggesting efficient interactions between the
molecule and the selected solvents. We further examined the
effect of medium polarity on the lifetime of 1e by collecting
measurements in W/i-PrOH mixtures at different ∆f (Figure 2f,
orange trace). Notably, the τ values show remarkable linearity
with ∆f up to 0.315, in good agreement with the brightness
behavior in the W/i-PrOH media (Figure 2c). These data confirm
that the fluorescence of 1e is regulated by a “classical”
solvatochromic behavior once the water content falls below a
threshold value (around 90% in mol). We also determined the
τ values of 1e upon 403-nm excitation in the same W/i-PrOH
mixtures in order to check the effect of the dual fluorescence
emission. Indeed, we always collected biexponential fluores-
cence decay curves except for the pure water case. To determine
the actual τ values of the high-energy emission component (τ470),
To determine the pattern of change in brightness between
water and organic media, we measured the water-relative
brightness of 1a and 1e in W/i-PrOH and W/Diox mixtures
characterized by different ∆f values (Figure 2c). For both
mixtures, the relative brightness of 1a displays a rapid increase
for ∆f above ∼0.31, whereas it follows a slower linear trend
for ∆f < ∼0.31. More precisely, the linear phase begins below
∆f ) 0.312 (10% mol) for W/i-PrOH or ∆f ) 0.309 (7%
mol) for W/Diox. Thermodynamic studies on W/IpOH or
W/Diox mixtures have shown that at low molar content of
organic solvent (5-10% in mol), water undergoes a transition
to structure II “clathrates”, which are characterized by an
ice-like arrangement of 136 H2O molecules to give 24
cavities. Eight of these cavities are large enough to accom-
modate guest molecules such as the organic solvent, whereas
the other are smaller and should contain “monomeric” H2O
molecules (i.e., not hydrogen-bonded with the infinite net-
work).59 At higher i-PrOH or Diox content, small clusters of
organic solvent and water molecules are formed.60 These
findings prompt us to attribute the sharp variation of brightness
of 1a to the structural rearrangement of water that leads
presumably to the loss of several H-bonds established with
surrounding H2O molecules. The linear trend at higher organic
content may be related to a more “classical” solvatochromic
effect.
Significantly, in both W/i-PrOH and W/Diox compound 1e
shows low and constant brightness in the same ∆f range where
its non-cyano counterpart 1a experiences a sharp increase of
emission. Below this region, brightness of 1e is linearly related
to polarity. This trend suggests that 1e cannot fluoresce in the
whole region of clathrate stability, becoming emissive only when
the water molecules become more “monomeric” and surrounded
by their organic counterparts. 1e is associated with a further
peculiar photophysical property: its dual fluorescence upon high-
energy excitation (below 405 nm) appears in all organic solvents,
and for water/organic mixtures, it is dependent on the organic
content (Figure 2d). In fact, both the high-energy (470 nm) and
the low-energy (540 nm) emission bands increase as the water
content of the mixture is decreased. The low-energy band (λ )
550 nm) is more sensitive to the polarity of the mixture than
its high-energy companion (λsh ) 470 nm). The latter one never
disappears, even in pure water. Thus, the ratio between the two
bands changes with the polarity of the mixture. This behavior
(59) Jerie, K.; Baranowski, A.; Koziol, S.; Glinski, J.; Burakowski, A.
Chem. Phys. 2005, 309, 277–282.
(61) Bizzarri, R.; Serresi, M.; Luin, S.; Beltram, F. Anal. Bioanal. Chem.
2009, 393, 1107–1122.
(60) Østergaard, K. K.; Tohidi, B.; Anderson, R.; Todd, A. C.; Danesh, A.
Ind. Eng. Chem. Res. 2002, 41, 2064–2068.
(62) Valeur, B. Molecular Fluorescence; Wiley-VCH Verlag GmbH:
Weinheim (DE), 2002.
9
1282 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Live Cell Imaging with Solvatochromic Coumarins
A R T I C L E S
Figure 3. (a) Confocal microscopy images of 1a in CHO cells (λexc: 403 nm). (b) (Left) Confocal microscopy image of 1e in CHO cells (λexc ) 403 nm,
λem ) 545-580 nm). (Right) Ratio of images collected at 545-580 nm and 440-480 nm (λexc ) 403 nm). (c) Fluorescence spectra of 1a and 1e in CHO
cells. (d) Lifetime-intensity color-coded map of 1a in CHO cells (λexc: 403 nm). (e) Color-coded lifetime map of the image reported in panel b (λexc ) 468
nm). (f) Histogram of lifetime distribution for 1a and 1e in CHO cells.
we fitted the biexponential decays imposing one lifetime equal
to that recovered by excitation at 468 nm. Importantly, we found
that λ470 is almost invariant with solvent polarity (Figure 2f,
blue trace) and its value (3.30 ns) is very close to its non-cyano
counterpart 1a.
2.6. Photophysical Properties in Cultured Cells. In order to
assess the photophysical behavior of the prepared coumarins
in biological environments, compounds 1a-e were examined
in living cells. The dyes were externally administered at low
concentration (0.5-1 µM) to Chinese Hamster Ovary (CHO)
cell cultures. After 15-30 min of exposition, cells were
imaged by a confocal microscope without removing the
external buffer by using 403 nm and/or 458 nm excitation
laser source (Figure 3a,b).
negligible fluorescence of 1a and 1e in water, as well as the
rather lipophilic properties of the dyes.
Remarkably, the emission profiles of the coumarins in the
subcellular region where they are active (Figure 3c) resemble
closely those determined during in Vitro experiments, indicating
that the biological environment does not affect the fluorescent
optical characteristics of the coumarins. The conserved dual
emission of 1e upon high energy (403 nm) excitation (Figure
3c) is particularly relevant in view of the use of this coumarin
as ratiometric emission indicator of polarity, working in the
visible region of the spectrum. A ratiometric plot of polarity
by 1e is reported in Figure 3b (right) and compared with the
lifetime map (Figure 3e) to which it is strictly related (see
below).
CHO cells remained viable for at least 2-3 h after addition
of the coumarins. No fluorescence was detected in the external
buffer, or in the nucleus, while collected images showed that
the coumarins are emissive in selected lipophilic subcellular
compartments. These findings are in agreement with the
The lifetimes of 1a and 1e were analyzed in cultured cells
by confocal FLIM. 1a was found to possess a quite uniform
monoexponential lifetime decay cell-wide (Figure 3d,f), in
excellent agreement with its constant lifetime observed in
medium of different polarity in Vitro. Also the intracellular
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1283

A R T I C L E S
Signore et al.
Figure 4. Colocalization of standard organelle markers (left, red channel) and 1a or 1e (right, green channel, and inset) in CHO cells. White spots represent
a selection of colocalized points (intensity ratio interval: 0.85-1.17). Colocalization results, obtained by a colocalization P-test according to Fay’s
method64 are reported in parenteses. (a) 1a (λexc: 403 nm, λem ) 420-600 nm) and membrane marker DiIC18(5)-DS (λexc: 633 nm, λem ) 650-750
nm) (poor colocalization, P ) 0.46); (b) 1a (λexc: 458 nm, λem ) 470-550 nm) and ER marker BODIPY TR glibenclamide (λexc: 561 nm, λem
570-680 nm) (strong colocalization, P > 0.95); (c) 1a (λexc: 403 nm, λem ) 410-535 nm) and lysosome marker Lysotracker (λexc: 561 nm, λem
)
)
570-700 nm) (moderate colocalization, P ) 0.68); (d) 1e (λexc: 458 nm, λem ) 470-600 nm) and membrane marker DiIC18(5)-DS (λexc: 633 nm, λem
) 650-750 nm) (strong colocalization, P > 0.95); (e) 1e (λexc: 458 nm, λem ) 470-600 nm) and ER marker BODIPY TR glibenclamide (λexc: 561
nm, λem ) 570-680 nm) (strong colocalization, P > 0.95); (f) 1e (λexc: 458 nm, λem ) 470-600 nm) and lysosome marker Lysotracker (λexc: 561 nm,
λem ) 570-700 nm) (strong colocalization, P > 0.95). Note that in experiments involving 1e the two dyes were imaged sequentially to avoid extensive
cross-talk.
lifetime behavior of 1e was in keeping with the properties
measured in Vitro. Upon 468 nm excitation, we detected
monoexponential decays with τ fairly dependent on the intra-
cellular location. Careful analysis showed that longer lifetimes
were associated with 1e localized in cell membranes or the
perinuclear region, owing to the more lipophilic character of
these regions (Figure 3e,f). Additionally, the lifetime map of
1e is very similar to its emission ratiometric map (Figure 3b,e),
i.e. there is a strict correspondence between longer fluorescence
lifetimes and increased low-/high-energy band ratio, as previ-
ously found in Vitro. These data highlight the possibility of using
1e as a gradual polarity indicator in bioenvironments by both
emission ratiometry and lifetime imaging.
Finally, the photostability of 1a and 1e, measured at high
laser power in living cells, was quite comparable with that of
YFP. Indeed, we measured fluorescence decrease with τ ) 39.5
( 2 s (93.1 kW/cm2), 13.7 ( 0.5 s (135 kW/cm2), and 30.3 (
0.8 s (106 kW/cm2) for YFP, 1a, and 1e, respectively.
2.7. Cellular Localization of Coumarins. To gain further
insight into the partitioning of 1a and 1e within different
subcellular domains, we performed some colocalization experi-
ments with standard organelle markers, in particular DiIC18(5)-
DS (membrane marker), BODIPY TR glibenclamide (ER
marker), and Lysotracker (lysosome marker). We found similar,
although not superimposable, localizations of 1a and 1e,
according to the different physicochemical properties of our
probes. 1a extensively accumulated in ER (Figure 4b), whereas
its localization in the cell membrane (Figure 4a), lysosomes
(Figure 4c), and Golgi apparatus (not shown) was rather limited.
1e accumulated in the cell membrane (Figure 4d) and ER (Figure
4e). Also, this coumarin is extensively sequestered in lysosomes
(Figure 4f), giving rise to punctate staining (see Figure 3b for
comparison). Cell localization differences can be clearly evi-
denced by a ratio plot of 1a and 1e fluorescence taken in cells
exposed to both coumarins (Figure 5a-c). These findings
suggest a preferential albeit different localization of 1a and 1e
in lipophilic compartments, in agreement with the hydrophobic
coumarin structure. Consistently, we measured log P values of
3.6 for 1a and 2.9 for 1e,63 substantiating the hypothesis of a
significant accumulation of the dye in the most apolar domains
of the cell but with different affinity. The presence of the
fluorophore in the cytosol could not be evidenced directly, due
to the nonemissivity of the solvatochromic probes in aqueous
environment. However, by means of simple fluorescence
recovery after photobleaching (FRAP) experiments, we found
that the coumarins have remarkably high diffusivity throughout
the cell (τ ) 3.1 ( 0.6 s, τ ) 5.6 ( 0.5 s for 1a and 1e,
(63) Taka´cs-Nova´k, K.; Avdeef, A. J. Pharm. Biomed. Anal. 1996, 14,
1405–1413.
(64) Fay, F. S.; Taneja, K. L.; Shenoy, S.; Lifshitz, L.; Singer, R. H. Exp.
Cell Res. 1997, 231, 27–37.
9
1284 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Live Cell Imaging with Solvatochromic Coumarins
A R T I C L E S
withdrawing groups onto aromatic rings is a common way to
promote solvatochromism. In such a case, the electronic
transition is often accompanied by a large transfer of electronic
charge from the electron donor toward the electron acceptor, a
phenomenon referred to as intramolecular charge transfer (ICT).
The significant electron redistribution between ground and
excited state upon photoexcitation results in a sudden change
of molecular dipole moment and therefore of stabilization by
the solvent. Nonetheless, a powerful solvatochromic effect is
not necessarily translated into a bright and polarity-dependent
fluorescence. Indeed, the addition of donating or withdrawing
group to otherwise fluorescent aromatic molecules may forbid
or severely hamper the fluorescence emission of the target
compound.
The coumarin aromatic structure is known to allow the
incorporation of electron donors and acceptors while retaining
most of its excellent fluorescent properties such as high quantum
yield and large Stokes’ shift. In this perspective, we targeted
the engineering of a novel toolbox of coumarins bearing different
functional groups in a donor-(coumarin core)-acceptor con-
figuration. The attention was mainly directed to alkylether
groups as good electron-donating functional units, since they
are amenable to further modification to yield chemical “hands”
for biomolecule binding. Additionally, alkylether groups prevent
the chemical reactivity problems (e.g., protonation, low yield
of further functionalization) that are often encountered when
the more popular electron-donating amino group is utilized. We
focused on benzothiazene and cyano functional units as acceptor
groups on account of their reported electron-withdrawing effect
on the physicochemical properties of conjugated aromatic
compounds.29 It is worth noting that all these groups were
singularly described to red-shift the basic coumarin spectra23
owing to a larger delocalization of the π orbital.
Our approach to the development of new solvatochromic
fluorescent probes followed a sequential scheme: (1) in silico
screening to identify interesting substitution patterns of the
coumarin structure from an optical viewpoint, (2) establishment
of a reliable and efficient synthetic procedure common for all
the target compounds (modular synthesis), (3) thorough in Vitro
analysis of the photophysical properties of the synthesized
probes, and (4) evaluation of their photophysical behavior in a
biological (e.g., cultured cell) environment. Overall, we sought
to set up a general and flexible scientific “platform” to engineer
probes tailored to different in ViVo applications.
Figure 5. (a,b) Confocal microscopy image of 1a (panel a, λexc ) 405
nm, λem ) 420-470 nm) and 1e (panel b, λexc ) 458 nm, λem ) 500-600
nm), concomitantly administered to CHO cells; (c) logarithm of the ratio
of images a and b to unveil different intracellular distributions of 1a and
1b; note that the log form was adopted to display spatial enrichment or
depletion in linear color-coded scale.
respectively),5 thus suggesting a partition equilibrium of the dyes
between the lipophilic compartments and the cytosol.
The computational DFT screening of coumarin structures was
carried out with the clear-cut goal of identifying donor-(coumarin
core)-acceptor patterns able to red-shift the absorption of pure
coumarin (310 nm) to the visible region of the electromagnetic
spectrum, while retaining a molecular structure amenable to
biochemical functionalization. Indeed, the visible region is by
far the preferred wavelength interval for the analysis of
biological specimens owing to the good compromise of high
spatial resolution and low photodamage to cellular structures.
Accordingly, most microscope apparatus for in ViVo imaging
are supplied with excitation sources emitting above 400 nm.
DFT/B3LYP analysis allowed for the identification of seven
interesting structures red-shifted of 70-180 nm compared to
base coumarin (Table 1, 1a-g). These structures are character-
ized by the presence of two alkylether- or one naphthyl-donor
group in positions 6-8, one benzothiazene group in position 3,
and for a subset, also by a cyano group in position 4. We noticed
that the nature of alkyl ether and the substitution regioselectivity
at positions 6-8 have little effect on absorption wavelength
3. Discussion
Environmentally sensitive fluorescent molecules appear ex-
tremely useful for in ViVo imaging applications. They can be
used as sensors to report on polarity changes in the surrounding
medium associated with relevant biochemical events such as
protein-protein recognition, membrane lipid restructuring, and
biomolecular self-assembly processes. Note that compared to
the more popular FRET-based sensors, solvatochromic probes
do not require two partners in close proximity and in well-
defined geometrical relationships to provide a signal which
monitors the target process at nanoscale. Hence, we recently
developed a renewed interest in molecular moieties optically
influenced by the polarity of their environment.
The solvatochromic behavior usually stems from large
variations of the molecular dipole upon photon absorption,
leading to different stabilization energies of the ground and
excited states by the solvent shell around the molecule. The
concomitant incorporation of electron-donating and electron-
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1285

A R T I C L E S
Signore et al.
(Table 1, 1a-c, 1e-g). Furthermore, DFT highlighted that the
concomitant incorporation of both benzothiazene and cyano
groups yields a synergistic red-shift effect on absorption (Table
1, 1e-g). These findings appeared promising in light of our
modular approach to probes tailored for specific applications
and prompted for the synthesis of 1a-g. Note that compound
1c represents already a “functional” evolution of 1b, as it can
be easily converted into a reactant for protein conjugation (we
shall describe this in a forthcoming paper).
Concerning the synthetic step, we set up a general procedure
based on a piperidine-catalyzed Knoevenagel reaction involving
substituted o-hydroxy benzaldehydes (Schemes 1-2, 3a,b′,c,d)
and a benzothiazene acetate derivative (Scheme 2, 2). The cyano
group was then added by a simple addition/elimination reaction.
Overall, the synthetic procedure allows obtaining the desired
coumarin in high yields just by selecting and preparing the
appropriate o-hydroxy benzaldehyde reactant. Note that the here-
adopted Knoevenagel reaction is known to be unaffected by
most functional groups that may be present on the reactants.
The absorption properties of the non-cyanocoumarins in
acetonitrile closely resemble those calculated by DFT. A good
agreement is seen also for cyano 1e, whereas DFT overestimated
the wavelengths of absorption peaks of 1f,g. We attribute the
latter discrepancy to larger distortion of the angle between the
two aromatic rings than predicted by the theoretical method.
Indeed, our calculation clearly correlates the twist angle with
the excitation energy. The absorption bands are in all cases broad
and often rich in vibrational substructure. Non-cyanocoumarins
can be excited efficiently by light sources in the 380-430 nm
range (Figure 1a-c); cyanocoumarins allow for excitation
wavelengths as long as 500 nm (Figure 1d). With the exception
of 1f-g, the prepared coumarins in acetonitrile showed broad
fluorescence emission characterized by large Stokes’ shifts
(Figure 1a-d) and high quantum yields (Table 2). Both
properties are extremely relevant in view of a future use of these
compounds for high-resolution imaging as they are at the basis
of large signal-to-noise ratio. 1e displays an interesting property:
when excited on the high-energy absorption shoulder below 400
nm it adds a minor emission band peaked at 470 nm to its
conventional band peaked at 570 nm. This dual fluorescence is
associated with a nearly constant band ratio for any excitation
below 400 nm (Supporting Information). We demonstrated,
however, that this effect is not the result of excimer formation
and must be related to the photophysics of the excited state.
From a solvatochromic viewpoint, we can split the fluorescent
coumarins into two subsets, one encompassing the non-cyano
derivatives, 1a-d, and the other cyano compound, 1e. The
analysis in solvents, or water/solvent mixtures, of different
polarities (measured by the orientation polarizability function
∆f) showed clearly that 1a-d are poorly or not emissive in
pure water, whereas they maintain approximately the same
quantum yield once the polarity falls below a certain value (∆f
≈ 0.31) no matter whether that happens in water/organic solvent
mixtures or pure protic/nonprotic organic solvents (Figure 2a,b).
Brightness enhancements ranged from 15- to 400-fold with
respect to pure water (Figure 2b). The sharp increase in
fluorescence is clearly related to a rearrangement of water
structure and the concomitant loss of strong H-bonding interac-
tions with the solute probe (Figure 2c).
Figure 6. Proposed model of electronic transitions in coumarins 1a-d (a)
and 1e (b).
simulations describe how water molecules interact with the
solute, during its ground-state dynamics, predicting the forma-
tion and stability of hydrogen bonds. By comparing the two
coumarins, we observed that only the oxygen atom at position
8 (O8) changes significantly its H-bonding degree between the
two molecules. In 1b we found ∼60% of full H-bonding
occupation (i.e., ∼60% of the molecular dynamics snapshots
show a configuration where a water oxygen is at a distance <3.5
Å and the O8-hydrogen-water oxygen angle is >150°),
whereas in 1c this value drops to ∼20%. Although these values
are just estimates of H-bonding capabilities, the comparison
between these two figures clearly indicates that O8 of 1b
establishes much stronger H-bonding interactions with water
molecules than O8 of 1c.
To further investigate the photophysical behavior of these
coumarins we determined the fluorescence lifetimes in water,
i-PrOH, and ACN of 1a and c (1b,d are practically nonfluo-
rescent in water, and reliable fluorescence decays could not be
obtained). Surprisingly, the lifetime is very similar in these three
solvents (Table 3). The radiative lifetime in water is more than
30 ns, absurdly 1 order of magnitude larger than that predicted
on pure theoretical basis from absorption/emission spectra. These
findings imply that the low quantum yield in water cannot be
related to an additional nonradiative decay channel from the
lowest vibrational level of the excited state (quenching mech-
anism). This means that, after the fast vertical transition due to
excitation, the coumarin can evolve along two different energy
landscapes: one is the usual internal conversion down to the
lowest vibrational level of the excited state, the other engages
a fast (ps) transition to an excited state of different nature
(possibly after some internal conversion) that provides a very
effective de-excitation pathway down to ground state (Figure
6). On account of recent results described for similar coumarin
structures,65 we are tempted to identify this second excited state
to a twisted intramolecular charge transfer (TICT) configuration.
DFT calculations show the electronic transition of 1a to be
associated with a significant rearrangement of electronic charge,
The H-bonding model helps explain also the different QY in
water of 1b and 1c in spite of the similar substitution pattern.
Indeed, we performed molecular dynamics simulations of these
two coumarins embedded in explicit water molecules. The
(65) Satpati, A. K.; Kumbhakar, M.; Nath, S.; Pal, H. Photochem. Photobiol.
2009, 85, 119–129.
9
1286 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Live Cell Imaging with Solvatochromic Coumarins
A R T I C L E S
in a typical ICT fashion. Usually ICT may evolve to TICT if
the interactions with the solvent molecules are favorable to the
latter configuration. Indeed, there is evidence for similar
coumarin systems in which a TICT state can be stabilized by
the establishment of strong H-bonding interactions with water
molecules. This would explain the high sensitivity of fluores-
cence emission of our compounds from the H-bonding capability
of the surrounding medium.
clearly the relative lipophilicity of different subcellular domains
(Figure 3e,f). Cells were found to be viable for at least 2-3 h,
indicating the low toxicity of the coumarins. It is worth noting,
however, that future bioapplications will involve much lower
concentrations and the conjugation will be with specific bio-
molecules so as to further reduce any toxic effect.
Even though future application of the dyes will involve
conjugation to biomolecules that will act as carriers for our
coumarins to specific domains, the peculiar staining showing
in the cells (Figure 3a,b) prompted us to achieve a better insight
in the distribution of our fluorescent probes. Colocalization with
standard organelle markers evidenced similar, albeit not identi-
cal, distribution in the cells. In particular, both 1a and 1e were
extensively sequestered in endoplasmic reticulum (Figure 4c,d);
additionally, 1e also stained more lipophilic domains such as
membranes and lysosomes (Figure 4b,f). This different subcel-
lular distribution can be evidenced by plotting the ratio of the
signal of the two dyes, concomitantly administered to the cells,
and it is in keeping with the different log P of the two probes.
The solvatochromic behavior of 1e is richer. First of all, the
quantum yield of emission shows good correlation with solvent
polarity (Figure 2a), and therefore, the brightness enhancement
relative to water is solvent dependent (Figure 2b). This
correlation is lost for pure water. More, 1e has constantly poor
emission in all of the polarity range where its non-cyano
counterpart 1a shows a sudden increase of brightness (Figure
2c). Below a certain polarity value (∆f ≈ 0.31), however, the
brightness of 1e is linearly related to ∆f, following a
“Lippert-Mataga” behavior. These findings suggest that 1e has
a solvent-dependent emission related to a differential stabiliza-
tion of ground and excited states below a certain polarity level,
and it is dark otherwise. Accordingly, the lifetime of 1e in water/
i-PrOH mixtures is strictly linear with ∆f except for pure water
(τ550, Figure 2f). The dual fluorescence of 1e upon <400 nm
excitation is conserved in all solvents. Nonetheless, the relative
ratio between the two emission bands is polarity-dependent
(Figure 2d). This behavior is particularly evident for water/
organic solvent mixtures and supports the use of 1e as a
ratiometric probe of polarity. Note that in water only the high-
energy emission band is present (τsh ≈ 470 nm, Figure 2d).
The dual fluorescence is reflected also in a biexponential (i.e.,
bicomponent) lifetime decay. The deconvolution of the decay
curves yielded the lifetime associated with the high-energy
emission band (τ470). Importantly, τ470 resulted independent of
medium polarity including also water. Thus, the high-energy
band of 1e looks very similar in wavelength and lifetime
behavior to the emission of its non-cyano parent 1a. These data
support a photophysical model encompassing two communicat-
ing excited states similar to that described for non-cyano
coumarins (Figure 6). The major difference, however, resides
in the existence of radiative channels from both excited states.
Additionally, the efficiency of the transition from the low-energy
excited state is polarity-dependent. Although we may be tempted
to attribute the two excited states to ICT and TICT in analogy
with non-cyanocoumarins, the complexity of the photophysical
pattern requires further investigation to provide a reliable
hypothesis, and this topic will be the subject of a forthcoming
paper.
4. Conclusions
In this work, we targeted the rational design and engineering
of a modular toolbox of polarity-sensitive probes to be used as
indicators of biochemical processes involving the change of
environmental polarity. One good example is the binding
between two biomolecules (e.g., two proteins) and the radical
variation of polarity at the interface of contact. In such cases,
polarity probes are thought to circumvent some of the critical
issues related to conventional FRET biosensors.
Our goal was accomplished by following a stepwise proce-
dure. First, we selected a photophysically promising chemical
structure, namely donor-(coumarin core)-acceptor where alky-
lethers or naphthyl played the role of electron donors and
benzothiazene ring and cyano the role of electron acceptors. In
silico screening allowed identifying good substituent configura-
tions for imaging in the visible region of spectrum. The selected
structures were synthesized in high yields by means of a flexible
and straightforward procedure that is insensitive to the nature
of the reactants. Hence, future functional derivatives will require
only an appropriate choice of the reactants. Most of the prepared
compounds showed good to excellent fluorescent properties for
imaging applications in terms of Stokes’ shift and brightness.
In terms of solvatochromism, we obtained two classes of
compounds. Coumarins devoid of a conjugated cyano group
were found to behave as polarity switches. They are almost
nonfluorescent in water or bioenvironments enriched in water,
but they sharply become very emissive once the polarity of the
environment falls below a certain level. On account of our
experimental data, we believe that this effect is associated with
a strong H-bonding stabilization by water molecules of a TICT
excited state, accessible from the original excited state, and
possessing a very efficient nonradiative decay channel. Pre-
liminary results yet to be published show that this effect is
maintained also when the coumarin is linked to a protein
structure.
A cyanocoumarin was found to exhibit a rich photophysical
behavior, very dependent on environment polarity. In par-
ticular, this compound is a promising concentration-
independent indicator of intracellular polarity on account of
its spectral ratiometric behavior when excited around 400
nm and its polarity-modulated mono-exponential lifetime
when excited above 450 nm.
Finally, we investigated the photophysical behavior and the
localization of our compounds in cultured cells. Our goals were
to verify that: (1) the compounds can be efficiently imaged in
a typical confocal microscope setup; (2) the photophysical
properties assessed in Vitro are not distorted by the biological
environment; (3) the compounds do not prevent cell viability;
(4) in cell environments enriched in water (e.g., cytoplasm) the
fluorescence is low. We found that all these requirements are
met. Indeed, after administration to cultured cells, the confocal
analysis showed that the coumarins stained lipophilic cellular
subdomains (Figure 3a,b, and 4). Conversely, our compounds
were almost nonfluorescent in cytoplasm, nucleoplasm, and
external buffer media. The photophysical properties in living
cells are completely superimposable on those in Vitro in terms
of emission spectra and lifetime decays (Figure 3c). In particular,
1e shows a polarity-dependent lifetime that allows identifying
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1287

A R T I C L E S
Signore et al.
discussions. This work was supported by INFM-CNR under the
framework of Seed Project “Solvatochromic fluorescent dyes for
bulk and single-molecule detection of intracellular native proteins”.
The partial support of the Italian Ministry for University and
Research (MiUR) under the framework of the FIRB project
RBLA03ER38 is also acknowledged.
Finally, tests of our compounds in cultured cells revealed that
they are very suitable for in ViVo high-resolution microscope
imaging in both spectral and lifetime modes. We believe that
these compounds, after appropriate functionalization and con-
jugation to target biomolecules, will represent remarkable tools
to investigate subtle biochemical processes in the cell environ-
ment. Furthermore, the photophysical comprehension of the
environmental sensitivity displayed by our coumarins will allow
for the engineering of further compounds tailored to specific
applications. Studies in both directions are currently under way.
Supporting Information Available: Detailed synthetic pro-
1
cedures; H and 13C NMR spectra for all compounds; spectro-
scopic data and high-resolution mass spectra for all compounds;
complete ref 53. This material is available free of charge via
the Internet at http://pubs.acs.org.
Acknowledgment. We gratefully acknowledge Prof. Fabio
Beltram for precious scientific suggestions during the research work
and in preparing this manuscript and Dr. Paolo Facci for stimulating
JA9050444
9
1288 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010