Synthesis of high contrast fluorescein-diethers for rapid bench-top sensing of palladium

Article abstract of DOI:10.1039/c5cc02192h

A new series of palladium sensors based on fluorescein bis(allyl ether) compounds has been designed. We utilized reduced fluoresceins as key synthetic intermediates. These probes exhibit negligible background fluorescence and rapid reaction with palladium, allowing a concentration of <100 ppt to be detected in minutes using a handheld UV lamp. This journal is

Full text of DOI:10.1039/c5cc02192h

ChemComm
View Article Online
View Journal
COMMUNICATION
Synthesis of high contrast fluorescein-diethers for
rapid bench-top sensing of palladium†
Cite this:DOI: 10.1039/c5cc02192h
Weston R. Kitley,a Peter J. Santa Maria,a Ryan A. Cloyda and Laura M. Wysocki*ab
Received 15th March 2015,
Accepted 13th April 2015
DOI: 10.1039/c5cc02192h
www.rsc.org/chemcomm
A new series of palladium sensors based on fluorescein bis(allyl ether)
compounds has been designed. We utilized reduced fluoresceins as
key synthetic intermediates. These probes exhibit negligible back-
ground fluorescence and rapid reaction with palladium, allowing a
concentration of o100 ppt to be detected in minutes using a
handheld UV lamp.
Fig. 1 The structure of the palladium sensor allyl Pittsburgh Green (1)
developed by Koide et al.5 and our proposed bis(allyl ether) 20,70-dichloro-
fluorescein palladium sensor 2.
Reactions involving palladium catalysis have found wide use in
modern synthetic chemistry due to their versatility, selectivity,
and relatively mild conditions.1 Although their ability to build
molecular complexity has led to increased utilization in the nonfluorescent (closed) lactone form. Although the open form
synthesis of Active Pharmaceutical Ingredients (APIs), palladium of fluorescein predominates in polar media, acylation or alkyla-
byproducts are often difficult to remove and toxic to ingest.2 tion of the phenolic oxygens can shift the equilibrium to the
Therefore, it is important to be able to efficiently and accurately closed form, rendering the molecule less fluorescent until removal
identify residual palladium in synthetic API targets destined for of these groups by a chemical reaction.7 For compound 1, the
use in biological studies, especially human consumption.
reduction of the key carboxyl group involved in the equilibrium
An attractive and innovative method to sense palladium involves and the presence of one allyl ether causes a significant decrease in
the installation of allyl groups onto phenolic fluorophores.3 Removal fluorescence. Although the open–closed equilibrium shifts dramati-
of these groups through a Pd(0)-catalyzed Tsuji–Trost deallylation4 cally, the alcohol–cyclic ether transition is still present in xanthene
releases a fluorescent molecule, giving a large increase in fluores- dye derivatives8 and some background signal exists due to the
cence. This was first demonstrated by Koide and coworkers using lingering presence of the open state. Upon specific reaction
the mono allyl ether Pittsburgh Green (1) (Fig. 1).5 Compound 1 has with palladium(0) and the resulting cleavage of the allyl ether,
been shown to be useful and selective for measuring low concentra- Pittsburgh Green fluoresces strongly.
tions of palladium using a fluorometer6 and is being developed for
To overcome the issue of background fluorescence, we
pharmaceutical applications by Koide and Welch.6d Nevertheless, planned to take advantage of the robust properties inherent
the current assay requires long reaction times (Bhours) and the in the open–closed equilibrium of fluorescein to achieve greater
probe exhibits significant fluorescence background, which precludes contrast and sensitivity to palladium(0). We proposed several
facile bench-top use of this probe.6
bis(allyl ether) fluorescein derivatives, based on the Pd(0) catalyzed
The relatively high background fluorescence of probe 1 can Tsuji–Trost deallylation reaction that has demonstrated selectivity
be explained by its structure. Fluorescein dyes exist in equili- with previous sensors. Koide and coworkers have shown that this
brium between a fluorescent (open) carboxylate form and a unmasking event is selective for palladium over other metals and it
takes place efficiently in the presence of several organic APIs.6b
a Department of Chemistry, Wabash College, 301 W. Wabash Ave., Crawfordsville,
We also incorporated substitution at the 20- and 70-positions of
the core dye, as in 2, anticipating an effect on the efficiency
of the unmasking reaction to reveal the fluorescent core. An
electron-withdrawing group at this position should increase the
ability of the xanthene core to act as a leaving group in the
key ether cleavage, thereby accelerating the unmasking event.
IN 47933, USA. E-mail: wysockil@wabash.edu
b Janelia Research Campus, Howard Hughes Medical Institute, 19700 Helix Dr.,
Ashburn, VA 20147, USA
† Electronic supplementary information (ESI) available: Reagents and instru-
ments, experimental procedures, additional spectroscopic data, 1H and 13C NMR
spectra. See DOI: 10.1039/c5cc02192h
This journal is ©The Royal Society of Chemistry 2015
Chem. Commun.

View Article Online
Communication
ChemComm
and sodium borohydride as a reducing agent to ensure conversion
of palladium(II) to the reactive palladium(0). Because the fluores-
cence of the unmasked fluorescein is pH dependent, basic condi-
tions were used to optimize the fluorescence intensity of the probe
after deallylation.11 Although previous studies required elevated
temperatures and aqueous phosphate buffer, our probes worked
efficiently at room temperature with methanol as the solvent and
6e
Pd(allyl)2Cl2 as the palladium source. These mild conditions
would be better suited for the rapid testing of trace amounts of
palladium in organic samples, like those that might be found in
the pharmaceutical industry, because of the solubility of a wide
variety of organic compounds in methanol.
The performance of the sensors synthesized here is sum-
marized in Fig. 2, which plots the palladium concentration
tested in each sample in ppb against the rate of fluorescence
increase, measured as the slope of the linear portion of the
Scheme 1 Synthesis of bis(allyl ether) fluorescein derivatives 7a–c.
For practical reasons, a faster rate of reaction results in a more kinetic fluorescence trace. This allows a direct comparison over
useful palladium sensor in most applications. a wide range of concentrations among sensors that have very
To synthesize fluorescein diether derivatives with the carboxyl different rates of reaction. Kinetic traces and intensity measure-
group intact, a protection strategy was employed,9 as shown in ments at a given time point for 7a–c can be found in the ESI.†
Scheme 1. For the synthesis of the first generation palladium The fluorescence changes observed upon addition of varying
sensor 7a, reduction of the central carbon of fluorescein diacetate concentrations of palladium to a sample solution of the sensor
(3a) under atmospheric pressure isolates the carboxylic acid for reveal a striking difference in rate of allyl ether cleavage when
further reaction. Subsequent protection via esterification using there is an electron-withdrawing substituent at the 20- and
2,4-dimethoxybenzyl (DMB) alcohol gives leuco-dye 4a in 99% 70-positions of the xanthene ring core. As expected, this sub-
yield over two steps. Selective hydrolysis of the acetate esters stitution makes the fluorescent dye a better leaving group,
reveals two phenolic groups that serve as nucleophiles in a increasing the sensitivity of the fluorogenic probe with signifi-
phase transfer reaction using allyl bromide (5) as the alkylating cantly shorter reaction times.
agent. This cleanly provides diether 6a in 62% yield over two
An interesting trend was observed when comparing com-
steps. Finally, mild oxidation using DDQ simultaneously pounds 7b and 7c. An explanation centered solely on electron-
removes the DMB group10 and returns the core xanthene dye withdrawing ability affecting rate of reaction might suggest that
to the proper oxidation state in 91% yield.
the 20,70-difluorofluorescein-based diether 7b would have
The synthetic approach described here results in selective the fastest deallylation due to fluorine’s electronegativity.
alkylation to form the desired diether derivative without the unwanted However, 7b and 7c are very close in their reaction rates, with
ether–ester byproduct. Additionally, this route features several benefits 20,70-dichlorofluorescein-based diether 7c exhibiting slightly
when considering the issues commonly encountered in the synthesis
of xanthene derivatives.9b Early reduction of fluorescein interrupts
the extended conjugation of the xanthene core. This isolated
phenolate moiety is a stronger nucleophile than the delocalized
fluorescein anion, allowing higher yields under mild alkylation
conditions. Each intermediate is soluble in common organic
solvents and purification can be done with silica gel column
chromatography. This compares favorably to a typical dye synthesis,
which can be limited in versatility and accessibility because it
requires high-boiling solvents like DMF and HPLC purification
of complex mixtures to obtain the final product.
The synthesis of the 20,70-dihalogenated bis(allyl ether)
fluorescein derivatives 7b and 7c followed the same straight-
forward strategy, shown in Scheme 1. Unsurprisingly, the hydro-
lysis of the acetate esters was more facile with the incorporation
of electron-withdrawing halogens, but each synthetic route
proved to be direct and high-yielding.
Fig. 2 Comparison of fluorogenic bis(allyl ether) fluorescein derivatives
7a–c. The slope is calculated from the linear portion of the kinetic trace
resulting from the addition of varying concentrations of palladium to a
With fluorogenic probes 7a–c in hand, their ability to sense
palladium was evaluated. Based on the optimized conditions
used by Koide and coworkers,6b we carried out the palladium
solution of 125 mM of the probe, 500 mM TFP, and 1.25 mM NaBH4 in
MeOH. Kinetic data is available in the ESI.†
assay in the presence of the ligand tri(2-furyl)phosphine (TFP)
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2015

View Article Online
ChemComm
Communication
faster kinetics. This experimental result matches the calculated
substituent constants (s) used for Hammett plots. Although
fluorine is stronger in its inductive effect compared to chlorine,
it is also more effective at electron-donation to the ring through
resonance. When both factors are taken into account, the
overall s value for chlorine is slightly higher.12 This observation
is important for the practical design of fluorogenic substrates
because the dichlorofluorescein derivative is more readily
available, but less frequently employed than the difluorofluor-
escein, or Oregon Green, counterpart. Our results demonstrate
that either derivative can improve the efficiency of reaction
dramatically over the unsubstituted fluorescein core.
Fig. 4 Sensor 7c in the presence of varying palladium concentrations:
(i) 1 ppb, (ii) 0.5 ppb, (iii) 0.25 ppb, (iv) 125 ppt, (v) 67.5 ppt, and (vi) no
palladium. After the palladium was added to a solution of 125 mM 7c,
500 mM TFP, and 1.25 mM NaBH4 in 2 mL MeOH, the mixture was stirred
for 5 min and illuminated by a handheld UV lamp for the photo.
After evaluating the efficiency of our palladium sensors, a
direct comparison can be made with the previously described
Pittsburgh Green sensor 1, as shown below in Fig. 3. Probe 1
was prepared by the published procedure5 and examined using
Koide’s optimized conditions for 1,6b which include stirring at present if any fluorescence is observed after a short reaction
45 1C in a 1.25 M phosphate buffered aqueous solution. time in ambient conditions. If multiple samples needed to be
Compound 7c was evaluated using the conditions described analyzed, this could easily be adapted to a microplate assay for
here. Fig. 3a shows the difference in fluorescence after stirring rapid determination of palladium concentration.6d
all samples for 5 min, in vials containing 1 with no palladium
The utility of this fluorescein diether design can be seen in
(i), 1 with 1 ppb palladium (ii), 7c with no palladium (iii), and Fig. 4, which shows a photograph of vials containing low concen-
7c with 1 ppb palladium (iv). It is clear that the background trations of palladium that have been stirred for 5 min and then
fluorescence of 1 in the absence of the analyte makes a visual illuminated by a handheld UV lamp. The vial on the right (vi)
comparison of the difference in fluorescence difficult (Fig. 3a, i has no palladium added, and therefore represents background
and ii). However, there is a marked difference in fluorescence fluorescence. Palladium in concentrations as low as 67.5 parts
using 7c due to the darkness of the sample without palladium per trillion (vial v) can be easily distinguished in this simple
(Fig. 3a, iii and iv). Fig. 3b shows a quantitative measurement of experiment. Although the calculated limits of detection for
what is seen in the photo, with 7c demonstrating a 314-fold these probes are 0.673 ppb for 7a, 0.455 ppb for 7b, and
increase in fluorescence after 5 min. After 1 h of stirring all 0.488 ppb for 7c (see ESI†), we found that the presence of lower
samples, a moderate increase in fluorescence of 1 occurs with concentrations can be visually identified. In a practical sense,
1 ppb palladium as reported previously,6 but background this represents an assay that could be performed quickly and
fluorescence would make visual detection of lower concentra- reliably by a chemist working at the bench interested in testing
tions inconclusive. Importantly, after 1 h, the sample contain- for trace palladium impurities in an organic sample.
ing 9b without palladium added is still nonfluorescent due to
the stability of the ether functionality (data in ESI†).
Harnessing the critical open–closed equilibrium of fluorescein in
the design of these molecules results in an easy-to-use and powerful
This experiment highlights the importance of low back- sensor for palladium. Utilization of the reduced dye intermediate 4
ground fluorescence in the implementation of useful fluoro- allows for an efficient, versatile synthetic strategy to obtain fluor-
genic probes. The conditions reported herein do not require escein diether compounds with low background fluorescence, such
elevated temperatures, high concentrations of buffers, or as 7a–c. For the detection of trace amounts of palladium, electron-
extended reaction times. Instead, a chemist could dissolve an withdrawing substituents speed up the unmasking reaction, but
organic sample in MeOH, add aliquots of the ligand, reducing the slower kinetics of the unsubstituted probe 7a may be ideal
agent, and 7c, and be confident that palladium impurities are for quantitatively monitoring higher concentrations of palladium
during the course of a reaction. With a colorless fluorogenic probe in
the absence of palladium, the presence of the analyte is easier to
detect in contrast. The diether structure of the derivatives reported
here traps the molecule in the nonfluorescent form and these
groups exhibit very slow hydrolysis, leading to a sustained dark
background, allowing for a larger dynamic range of use. Future
work may be done to anchor these probes to a solid support,
thereby providing a further-simplified test for the presence of
palladium in practical applications.
This work was supported by Howard Hughes Medical Institute,
Wabash College, and a contribution from Lilly. We acknowledge
Fig. 3 Direct comparison between the assay described by Koide et al.
with sensor 1 and the reported sensor 7c. (a) After 5 min stirring times, the
vials contain: (i) no palladium with 1, (ii) 1 ppb palladium with 1, (iii) no
Luke D. Lavis and Jonathan B. Grimm (Janelia Research Campus)
for contributive discussions.
palladium with 7c, and (iv) 1 ppb palladium with 7c. (b) Fluorescence
measurements of each experiment in (a) at 5 min time point.
This journal is ©The Royal Society of Chemistry 2015
Chem. Commun.

View Article Online
Communication
ChemComm
(e) J. Li, J. Yu, J. Zhao, J. Wang, S. Zheng, S. Lin, L. Chen,
M. Yang, J. Shang, X. Zhang and P. R. Chen, Nat. Chem.,
2014, 6, 352; ( f ) H. Li, J. Fan and X. Peng, Chem. Soc. Rev., 2013,
42, 7943.
7 L. D. Lavis and R. T. Raines, ACS Chem. Biol., 2008, 3, 142.
8 S. Kenmoku, Y. Urano, H. Kojima and T. Nagano, J. Am. Chem. Soc.,
2007, 129, 7313.
9 (a) L. M. Wysocki, J. B. Grimm, A. N. Tkachuk, T. A. Brown, E. Betzig
and L. D. Lavis, Angew. Chem., Int. Ed., 2011, 17, 431; (b) L. D. Lavis
and R. T. Raines, ACS Chem. Biol., 2014, 9, 855.
10 C. Kim and P. Misco, Tetrahedron Lett., 1985, 26, 2027.
11 N. O. Mchedlov-Petrossyan, M. I. Rubtsov and L. L. Lukatskaya,
Dyes Pigm., 1992, 18, 179.
Notes and references
1 J. Magano and J. R. Dunetz, Chem. Rev., 2011, 111, 2177.
2 C. E. Garrett and K. Prasad, Adv. Synth. Catal., 2004, 346, 889.
3 (a) J. Chan, S. C. Dodani and C. J. Chang, Nat. Chem., 2012, 4, 973;
(b) H. Zheng, X.-Q. Zhan, Q.-N. Bian and X.-J. Zhang, Chem. Commun.,
2013, 49, 429; (c) X. Wang, Z. Guo, S. Zhu, H. Tian and W. Zhu, Chem.
Commun., 2014, 50, 13525.
4 B. M. Trost and T. J. Fullerton, J. Am. Chem. Soc., 1973, 95, 292.
5 F. Song, A. L. Garner and K. Koide, J. Am. Chem. Soc., 2007, 129, 12354.
6 (a) A. L. Garner, F. Song and K. Koide, J. Am. Chem. Soc., 2009,
131, 5163; (b) F. Song, E. J. Carder, C. C. Kohler and K. Koide,
Chem. – Eur. J., 2010, 16, 13500; (c) D. Li, L. D. Campbell, B. A. Austin
and K. Koide, ChemPlusChem, 2012, 77, 281; (d) X. Bu, K. Koide,
E. J. Carder and C. J. Welch, Org. Process Res. Dev., 2013, 17, 108;
12 J. Bromilow, R. T. C. Brownlee, V. O. Lopez and R. W. Taft, J. Org.
Chem., 1979, 44, 4766.
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2015