Design and synthesis of epicocconone analogues with improved fluorescence properties

Article abstract of DOI:10.1021/ja506914p

Epicocconone is a natural latent fluorophore that is widely used in biotechnology because of its large Stokes shift and lack of fluorescence in its unconjugated state. However, the low photostability and quantum yields of epicocconone have limited its wid

Full text of DOI:10.1021/ja506914p

Article
pubs.acs.org/JACS
Design and Synthesis of Epicocconone Analogues with Improved
Fluorescence Properties
Philippe A. Peixoto,† Agathe Boulange,† Malcolm Ball,‡ Bertrand Naudin,§ Thibault Alle,† Pascal Cosette,§
́
Peter Karuso,*,‡ and Xavier Franck*,†
†Normandie Universite,
Mont-Saint-Aignan Cedex, France
‡Department of Chemistry and Biomolecular Sciences, Macquarie University, Sydney, NSW 2109, Australia
§CNRS UMR 6270, PBS & FR3038, Plateforme Proteo
mique PISSARO, Universite de Rouen, 76821 Mont-Saint-Aignan, France
́ ́ ̀
COBRA, UMR 6014 & FR 3038; Universite de Rouen; INSA Rouen; CNRS, 1 rue Tesniere, 76821
́
́
S
* Supporting Information
ABSTRACT: Epicocconone is a natural latent fluorophore that is widely used in biotechnology because of its large Stokes shift
and lack of fluorescence in its unconjugated state. However, the low photostability and quantum yields of epicocconone have
limited its wider use, and in the absence of a total synthesis, this limitation has been a long-standing problem. Here we report a
general strategy for the synthesis of epicocconone analogues that relies on a 2-iodoxybenzoic acid-mediated dearomatization and
on the replacement of the triene tail of the natural product by an aromatic ring. This design element is general and the synthesis
is straightforward, providing ready access to libraries of polyfunctional fluorophores with long Stokes shifts based on the
epicocconone core. Our structural modifications resulted in analogues with increased photostability and quantum yields
compared with the natural product. Staining proteomic gels with these new analogues showed significant lowering of the
detection limit and a 30% increase in the number of low-abundance proteins detected. These epiccoconone analogues will
substantially improve the discovery rate of biomarker needles in the proteomic haystack.
The recent discovery of the natural product epicocconone
(1) from the fungus Epicoccum nigrum5 revealed the first
reversible-covalent latent fluorophore, which has found wide
biotechnological applications in Western blotting, proteomic
gel staining, protein quantification, live-cell imaging, and
monitoring of enzymatic activity.6 Mechanism of action studies
revealed that the masked aldehyde of epicocconone reacts with
amine residues (e.g., lysine residues in proteins) to create a red-
fluorescent enamine that forms an internal charge transfer
complex that is stable at pH 2.4 (Scheme 1) and highly
emissive in the red (610 nm).7
INTRODUCTION
■
The power of fluorescent probes can be enhanced by
controlling how and when the probe becomes excitable. Such
latent fluorophores result in high specificity and low back-
ground and thus high sensitivity. Most latent fluorophores
either react covalently with their target1 or work by the removal
of a protecting group2 or quenching group3 and are thus
irreversible. However, random covalent modification of target
biomolecules is generally incompatible with proteomics (where
protein identification relies on mass spectrometry of peptides
released by tryptic digestion) unless considerable effort is
expended in modifying the bioinformatics software.4 This is
especially true of small molecules that modify lysine residues, as
these are no longer recognized by trypsin. Trypsin digestion of
single proteins or proteomes is the stalwart of protein
identification by mass spectrometry. Covalent modification of
biomolecules is also of limited use in live-cell applications,
where covalent modification by xenobiotics is often toxic.
Epicocconone is unique in that at neutral pH the fluorophore
reverts back to the nonfluorescent masked aldehyde in a base-
catalyzed reaction (Scheme 1), providing traceless detection of
proteins in a variety of formats.7,8 The compound is also freely
Received: July 9, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
advantages of 1 but have improved photophysical properties
such as increased quantum yield and photostability.
Scheme 1. Mechanism of the Reversible Reaction of
Epicocconone (1) with Proteins To Produce a Highly
Fluorescent Enamine
RESULTS AND DISCUSSION
■
The design of more potent analogues of 1 hinged on replacing
the triene tail with an isosteric aromatic ring and the ring-C
alcohol with a dimethyl group (Scheme 3; shown in green and
Scheme 3. Retrosynthetic Analysis
water-soluble yet cell-permeable and has a large Stokes shift.
Few small molecules have been found that emit in the red upon
irradiation in the UV or blue, even though such a property has
been identified as having great importance in dual staining
methods.9 Compounds that fall into this class include dapoxyl-
(2-aminoethyl)sulfonamide, Fura Red, Sypro Ruby, and di-8-
ANEPPS (Scheme 2), but none of these combine a large Stokes
Scheme 2. Examples of Dyes Excited in the Near-UV/Blue
and Emitting in the Orange/Red Region
blue, respectively). These modifications should have no effect
on the fluorescent scaffold but avoid photoisomerization or
photooxidations that could lead to decomposition, thereby
reducing the photobleaching observed for the natural product.
It has also been suggested that photoisomerization of the triene
tail is the major nonradiative decay mechanism for 1 in
solution, leading to a low quantum yield.8a Furthermore, the
geminal disubstitution on the C ring should reduce the rate of
recyclization to the masked aldehyde by replacing the 1,2-diol
(shown in blue in Scheme 1) with a less nucleophilic (more
hindered) tertiary alcohol.
Retrosynthetically, analogues of epicocconone could be won
from simple starting materials such as 2-methylresorcinol
dimethyl ether (2) or methyl atratate (3) through three key
steps (Scheme 3). The tricyclic core of epicocconone 4 could
be synthesized by trapping of an acyl ketene by 5 with
concomitant intramolecular Knoevenagel condensation in a
one-pot procedure (step C). In turn, 5 could be envisaged as an
oxidation product of 6 (step B), which in turn could be readily
prepared from either commercially available 2 or 3 by alkylation
(step A or A′).
shift with the “turn on”/“turn off” capacity of epicocconone.10
A large Stokes shift is advantageous because it minimizes
interference from intrinsic fluorescence and Rayleigh scattering
as well as self-quenching when two or more fluorophores are
within the Forster radius. However, epicocconone suffers from
̈
photobleaching when used, for example, over a transilluminator
for manual spot picking in 2D gel electrophoresis.11 This may
be due to photooxidation/photoisomerization12 of the triene
tail or photooxidation of the alcohols.13
As proteomes are highly complex with a large dynamic range
of protein concentration, it is imperative that new tools be
developed for identifying and quantifying low-abundance
proteins because of their pharmaceutical, diagnostic, and
medical value.14 Methods that allow deeper interrogation of a
proteome than current methods are thus in high demand, as all
of the new proteins seen are, by definition, low-abundance
proteins. This challenge inspired us to investigate the design
and synthesis of new fluorophores that encompass the
Two strategies for the synthesis of lactone 6 were successful
(Scheme 4). Our first strategy started with the formation of
amide 7 by the reaction of 2 with iPrNCO in the presence of
AlCl3. Directed ortho metalation of amide 7 using 2 equiv of
tert-butyllithium complexed in situ with tetramethylethylenedi-
amine (TMEDA) and subsequent reaction of the transient
aryllithium species with 2,2-dimethyloxirane produced 8 and
introduced the gem-dimethyl group. Lactonization of amido
alcohol 8 was achieved by refluxing in toluene with
B
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
toluene, but all other solvents led to over-reduction and
isolation of mixtures containing the diol. Treatment of
hemiacetal 10 with the λ5-iodinane derivative 2-iodoxybenzoic
acid (IBX) in the presence of trifluoroacetic acid (TFA) and
water deprotected the MOM ethers, dehydrated the hemiacetal
to the oxonium ion (as observed by NMR), and effected ortho
dearomatization.15 The addition of water was essential in this
reaction as it increased the yield and rate of reaction and
prevented the formation of TFA esters.16 Under these
conditions, the second key intermediate 5 was obtained in an
overall yield of 68% from 6.
Scheme 4. Synthesis of 6 via Two Different Routes
Analogues of 1 bearing a ketone side chain could be
synthesized from 5 by reaction with an acyl ketene.17 The
intermediate β-keto esters could then undergo Knoevenagel
condensations through 11 or 11′ to yield either 6,7-furanone
12 or 7,8-furanone 13, respectively (Scheme 6).
a
Reagents and conditions: (a) iPrNCO (1.25 equiv), AlCl3 (1.05
equiv), CH2Cl2, 0 °C, 3 h, 95%. (b) TMEDA (2.0 equiv), t-BuLi (2.2
equiv), THF, −15 °C, 30 min, then 2,2-dimethyloxirane (2.2 equiv),
25 °C, 1.5 h, 92%. (c) CSA (1.2 equiv), toluene, reflux, 2 h, 76%. (d)
AlCl3 (10.0 equiv), CH2Cl2, reflux, 20 h, 99%. (e) NaH (2.0 equiv),
MOM-Cl (5.0 equiv), THF, 25 °C, 2 h, 100%. (f) NaH (3.0 equiv),
MOM-Cl (3.0 equiv), THF, 25 °C, 3 h, 87%. (g) DIPA (1.0 equiv), n-
BuLi (1.0 equiv), THF, −78 °C, 30 min, then acetone (0.8 equiv), 5
min, 60%.
Scheme 6. Reaction of 5 with in Situ-Generated Acyl
Ketenes Followed by Knoevenagel Cyclization
camphorsulfonic acid (CSA). Finally the phenol protecting
group was swapped from methyl to methoxymethylene
(MOM) to yield 6 in two steps. The second, shorter strategy
started with MOM protection of 3 before regioselective
formation of the benzylic anion of 9 by treatment with lithium
diisopropylamide (LDA) at low temperature. The gem-dimethyl
group was then introduced by the addition of freshly distilled
acetone to the anion. While the first longer route (five steps)
gave a high overall yield of 66%, the shorter second route was
only two steps but offered a slightly lower overall yield of 52%.
Both routes are quite practical and suitable for multigram-scale
synthesis of 6.
The next key intermediate 5 (Scheme 5) was obtained by
first reducing 6 to hemiacetal 10 with diisobutylaluminum
hydride (DIBAL). The reaction proceeded quantitatively in
The one-pot installation of the acyl furanone could be
envisaged through trapping of an acyl ketene formed from
thermal fragmentation of dioxin-4-ones 14 by the alcohol of
5.16a,18 A library of analogues could be synthesized through this
route simply by starting with different carboxylic acids (Scheme
7). In practice, the synthesis of 14 proceeded smoothly via
condensation of the lithium enolate of tert-butyl acetate with
Weinreb amides. The reaction of the resulting β-keto esters
with acetone in acidic conditions yielded the dioxinones in 71−
82% yield (Scheme 7). The reaction of 14a with 5 at 110 °C in
the presence of triethylamine gave 6,7-furanone 12a in 86%
yield with no isolatable amount of the undesired 7,8-
regioisomer 13a arising from path b in Scheme 6. Starting
with different dioxinones, compounds 12a−d were obtained.
However, when R was an alkyl chain (e.g., 14c), mixtures were
obtained with the undesired regioisomer 13c being the major
product. This unexpected reversal of regioselectivity warranted
further investigation.
a
Scheme 5. Synthesis of 5 by Oxidative Dearomatization
After the reaction of the acyl ketenes with 5, for cyclization to
the acyl furanone to occur, the enolized side chain must adopt a
suitable conformation that allows cyclization with concomitant
proton transfer (Scheme 6). Two possible transition states
a
Reagents and conditions: (a) DIBAL (1.3 equiv), toluene, −78 °C,
20 min. (b) TFA (7.0 equiv), water (20 equiv), IBX (2.0 equiv),
CH2Cl2, 25 °C, 3 h, 68% over two steps.
C
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 7. Synthesis of Epicocconone Analogues from 5 and
Dioxinones 14
Table 1. Thermodynamic Properties of 11 and 11′ Based on
a
High-Level DFT Calculations
a
Gibbs free energy
(kJ/mol)
entropic energy
(kJ/mol)
Boltzmann
b
R
distribution
11
Ph
Ph
Et
739.91
742.06
689.37
688.69
306.89
304.10
285.71
0.66
0.33
0.45
0.55
11′
11
11′
Et
286.35
a
b
At 383 K using the Turbomole freeh package. Based on differences
in the Gibbs free energies of 11 and 11′.
R is Ph, 11 has a higher entropic energy (ΔTΔS = 2.8 kJ/mol)
while when R is Et, 11′ has the higher entropy (ΔTΔS = 0.6
kJ/mol) relative to 11 at 383 K.
In terms of spectral characteristics, compounds 12a−d have
low extinction coefficients (2000−4000 M−1 cm−1) and low
quantum yields (Table 2), and their excitation and emission
curves are blue-shifted by ∼20 nm compared with those of
epicocconone.8a Fluorescence data were obtained in 0.1% SDS
at pH 2.5 to mimic the environment in an SDS-PAGE
electrophoresis gel after fixing. While analogues 12a−d did
react with protein in water to yield a fluorescent enamine, the
quantum yield increased only slightly. Analogue 12c, with an
aliphatic side chain, is ∼10 times less fluorescent than 12a, and
its regioisomer 13c is completely devoid of any fluorescence.
While no useful probes arose from this work, it was now clear
that the 6,7-furanone and the β-diketone were critical design
features that needed to be incorporated to produce useful new
fluorophores. In order to access analogues incorporating the β-
diketone side chain, as in the natural product, we designed
dioxin-4-ones 15, which could be prepared by condensation of
the lithium enolate of commercially available 2,2,6-trimethyl-
4H-1,3-dioxin-4-one with various acyl chlorides (Scheme 8).
Unfortunately, thermal fragmentation of 15d under the same
basic reaction conditions as with dioxin-4-ones 14 (vide supra)
resulted in none of the desired product 4d, presumably because
of intramolecular trapping of the ketene to form 2-pyranones.
The solution to this problem was simply to add Et3N only after
ketene formation and trapping with 5. Even though the level of
conversion for this reaction was quite high (∼70% by NMR)
the isolated yield was moderated because of partial decom-
position on silica gel. Again, only the desired 6,7-furanones 4a−
d bearing aryl moieties and enolized 1,3-diketones were
obtained (Scheme 8). Methylated dioxinones 16 were prepared
from 15 using methyl iodide followed by protection of the aryl
ketone as the silyl ether. Reaction with 5 yielded 17 with a non-
enolized 1,3-diketone in the same moderate yields (Scheme 8).
Although the isolated yields of 4 and 17 were moderate, this
one-pot procedure performed four steps in one: fragmentation
of β-keto dioxin-4-one 15 or 16 to a putative β-keto acyl
ketene, acylation of 5, cyclization, and finally dehydration
exclusively to give the desired 6,7-furanone skeleton of
epicocconone. The yields for this step could not be improved
because β-keto acyl ketenes are unstable intermediates that are
prone to cyclize to the 6-aryl-4-hydroxy-2H-pyran-2-ones or
react with traces of water leading to β-keto acids, which easily
decarboxylate.14b However, this method is step-economical,
general, and achievable on a gram scale.
a
Reagents and conditions: (a) (COCl)2 (1.25 equiv), CH2Cl2/DMF
(cat.), 1.5 h; solvent removal, HN(Me)OMe (1.4 equiv) then Et3N
(1.0 equiv), CH2Cl2, 2 h, 86−96% yield over two steps. (b) DIPA (2.0
equiv), BuLi (2.0 equiv)/THF, −78 °C, then t-BuOAc (3 equiv), 1 h,
91−99% yield. (c) Ac2O (15 equiv), H2SO4 (1 equiv)/acetone, 0−25
°C over 45 min, 71−82% yield. (d) 5 (0.67 equiv), Et3N (1.33 equiv),
molecular sieves (4 Å; 200 mg/mmol)/toluene, 110 °C, 3 h.
could be envisaged, one leading to the 6,7-furanone (path a)
and the other to the 7,8-furanone (path b). The steric
requirements, if R is aryl, are greater for path b, as in the
transition state (TS) the R group lies over the pyran ring and
interacts directly with the gem-dimethyl group. However, if R is
alkyl, this steric hindrance is reduced because of the flexibility
and smaller size of the side chain compared with an aryl group.
In both cases, a concerted mechanism must produce the all-cis
γ-lactone (11 or 11′), which is set up for a facile syn
elimination of water to give the respective furanone (12 or 13).
This was confirmed by molecular modeling (Figure S1 in the
Supporting Information). This qualitative model predicts a
preference for 12 over 13 if R is aryl and less of a preference if
R is alkyl. This was what was observed, except that with alkyl
groups the undesired 7,8-furanone 13 was the major product,
suggesting that other forces were at play.
To investigate this further, we performed density functional
theory (DFT) calculations (DFT-D3//BP-86/TZVPP)19 using
a continuum solvent model (COSMO)20 for toluene on the
intermediate all-cis γ-lactones (11 and 11′; R = Et and Ph).
Under the cyclization conditions (toluene, 110 °C), one would
expect the thermodynamic products to dominate if the
intramolecular Knoevenagel condensation is reversible
(Scheme 6). Frequency calculations on the geometry-optimized
intermediates (11 and 11′) confirmed that all of the predicted
conformations were stationary points, and thermochemical
properties were calculated at 383 K using Turbomole (Table
1). Gibbs free energy calculations predicted that the 12d:13d
ratio would be 1:0.5 while the 12c:13c ratio would be 0.8:1.
The latter is close to the experimental result (Scheme 7), while
the former predicts the correct major product but the reaction
is actually more selective for the desired regioisomer, suggesting
that TS influences (Scheme 6 and Figure S1 in the Supporting
Information) may also play a role in the observed product
distribution. The differences in Gibbs free energy are
predominately due to differences in entropy (Table 1). When
With several analogues in hand, we set out to characterize
their fluorescence behavior (Figure 1 and Table 2). In the case
of analogues 17a and 17d, the non-enolized 1,3-diketones were
obtained and had emission wavelengths similar to those of
D
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
fluorescent in water (ΦF ≈ 40%) than in the presence of
protein (ΦF ≈ 17%).
Scheme 8. Synthesis of Epicocconone Analogues from 5 and
Dioxinones 15 and 16
a
Interestingly, while all of the compounds except 13 were
fluorescent and displayed one emission line, most displayed
three distinct excitation maxima corresponding to S0−S3, S0−
S2, and S0−S1 transitions (Table 2).
The advantage of 1 in protein detection is that it is virtually
nonfluorescent in water (ΦF < 1%), a characteristic shared with
4a and 4d.22 However, in 0.1% SDS solution (Table 2), the
quantum yields increase by an order of magnitude. This is
presumably due to rigidification of the fluorophore in the
viscous environment around SDS clusters and/or a reduction in
collisional quenching from water. In contrast, 4b and 4c have
considerably higher fluorescence than 1 in water22 and in 0.1%
SDS (Figure 1A).
The pyridyl analogue (4c) that seemed quite promising
produced an enamine with bovine serum albumin (BSA) that
was much less fluorescent than that of 1. Compounds 4a and
4d were also less fluorescent than 1 in the presence of excess
protein (Figure 1B), and this translated into inferior protein
stains (Figure S7 in the Supporting Information). In the
presence of excess protein, only 4b matched 1 in fluorescence
performance (Figure 1B), and it was selected for further study.
The application of 4a−d for staining of proteins in
electrophoresis gels was tested using 1D gels (Figure S7 in
the Supporting Information) and 2D gels (Figure S9 in the
Supporting Information), and the results confirmed that
compound 4b had the best protein staining characteristics.
We chose to analyze the performance of 4b in more detail by
comparison with (1) Deep Purple Total Protein Stain23
because Deep Purple contains 1 as its active ingredient and
(2) Sypro Ruby24 because it is the most popular fluorescent gel
stain but is based on very different chemistry (Scheme 2).
Analysis of serial dilutions of standard proteins (Figure 2)
indicated that 4b gave much darker staining and was generally
twice as sensitive as the commercial products in 1D gels (Table
3 and Figure S6 in the Supporting Information), achieving
picogram (low femtomolar) sensitivity. As both Deep Purple
and 4b are end-point stains, this difference is not related to
staining times or protocols. We used the standard Deep Purple
protocol in both cases and made no attempt to optimize the
protocol for 4b.
a
Reagents and conditions: (a) DIPA (2.0 equiv), n-BuLi (2.0 equiv),
THF, −78 °C, 30 min, then ArCOCl (1.0 equiv), −40 °C, 2 h, 41−
62% yield. (b) 5 (0.67 equiv), molecular sieves (4 Å; 200 mg/mmol),
toluene, 110 °C, 20 min, then Et3N (1.33 equiv), 1 h, 27−35% yield.
(c) K2CO3 (5.0 equiv), MeI (2.0 equiv), acetone, reflux, 2 h. (d)
TBSOTf (1.5 equiv), DIEA (3.0 equiv), CH2Cl2, 25 °C, 2 h, 99% yield
over two steps. (e) 5 (0.67 equiv), Et3N (1.33 equiv), molecular sieves
(4 Å; 200 mg/mmol), toluene, 110 °C, 3 h, 25−29%.
12a−d (i.e., blue-shifted by ∼20 nm compared with 1) and low
quantum yields. This indicated that the second carbonyl in 17
is not part of the chromophore. In contrast, analogues 4a−d
were obtained in the keto−enol form (from NMR) and
exhibited excitation and emission spectra very similar to those
of 1, confirming that the triene tail is not required for strong
fluorescence but that the β-keto−enol is.
In most cases, the formation of the enamines of 4 (with
protein) resulted in dramatic increases in fluorescence and
concomitant red shifts in the excitation and emission
frequencies arising from increased conjugation, higher dipole
moment in the excited state (Figure 1), and an increased charge
transfer distance compared with the first series of analogues
12a−d.21 The exception to this was 4c, which was more
Similar results were obtained in the more critical 2D gel
electrophoresis (Figure 3), where 4b was able to reveal 33%
Figure 1. Emission spectra of epicocconone (1) and analogues 4a−d in water at pH 2.5 with 0.1% SDS in the (A) absence or (B) presence of excess
protein (BSA). All spectra are normalized on identical concentrations of fluorophores (10 μM) and for excitation at the longest λex (Table 2).
E
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 2. Relative Quantum Yields, Molar Extinction Coefficients, and Brightnesses of Epicocconone (1) and Analogues 4a−d,
12a−d, and 17a−d in H2O (pH 2.5) with 0.1% SDS in the Absence or Presence of Excess Protein (BSA)
without BSA
with BSA
c
c
λem
ε
brightness
λem
ε
brightness
a
b
d
e
compd
λex (nm)
(nm) (M−1 cm−1
)
ΦF (%)
(M−1 cm−1
5.08
)
λex (nm)
(nm) (M−1 cm−1
)
ΦF (%)
(M−1 cm−1
4.51
)
1
258, 319,
444
535
538
538
542
534
519
520
526
517
518
68600
17600
39200
11300
14500
4000
7.4 0.2
4.8 0.2
19.8 0.4
40.3 0.2
27.0 0.4
2.8 0.2
3.0 0.1
13.4 0.2
34.6 0.7
40.1 1.0
256, 370,
518
611
605
613
614
607
589
587
589
586
587
26400
11600
18400
5500
10800
700
17.1 0.2
21.7 0.5
21.2 0.7
16.7 0.5
18.8 0.5
4.5 0.1
5.8 0.2
2.8 0.1
3.5 0.1
3.0 0.1
4a
253, 303,
420
0.85
7.76
4.55
3.92
0.11
0.07
0.27
3.05
5.29
258, 369,
509
2.52
3.90
0.91
2.03
0.03
0.02
0.01
0.18
0.08
4b
252, 308,
435
255, 372,
518
4c
252, 312,
425
254, 370,
527
4d
254, 303,
417
257, 370,
518
12a
12b
12d
17a
17d
260, 320,
383
257, 345,
480
262, 317,
381
2200
247, 347,
477
300
308, 383
2000
260, 352,
477
300
279, 391
8800
271, 350,
485
5100
2600
248, 322,
392
13200
250, 353,
492
a
b
Excitation spectrum recorded with emission at 520 nm. Quantum yield relative to Lucifer Yellow CH (ΦF = 0.21) standard deviation of three
c
d
e
readings. Brightness is defined as ΦF × ε/1000. Excitation spectrum recorded with emission at 620 nm. Relative to rhodamine 6G (ΦF = 0.76);
see Figure S4 in the Supporting Information.
Figure 2. 1D gels stained with (A) Deep Purple and (B) compound
4b run and scanned under identical conditions with similar
concentrations of dye (∼1 μg/mL in each case).
Table 3. Limits of Detection (pg) for Analogues 4a−d and
Deep Purple (Epicocconone 1) in 1D Gel Electrophoresis
1
4a
4b
4c
4d
phosphorylase B
albumin
1838
917
2562
1301
668
799
560
763
593
786
1534
1005
1210
864
1393
1123
866
ovalbumin
1766
877
carbonic anhydrase
soybean trypsin inhibitor
838
655
1338
1763
401
893
more spots than Deep Purple (Table S3 in the Supporting
Information) under the same illumination conditions, allowing
a much deeper analysis of the proteome. In 2D gels, Deep
Purple performs well on low-molecular-weight and basic
proteins. Analogue 4b had similar characteristics but resulted
in better staining of acidic proteins (left-hand side of the gel).
However, in general the staining characteristics of 4b and 1
were very similar, as expected.
In comparison to Sypro Ruby, 4b had substantially lower
levels of noise, better peak shapes, and less speckling (Figure
4). The staining profiles were quite similar for the two dyes, in
accordance with the fixation targets of the fluorescent probes
(lysines for epicocconone and its analogues and basic residues
Figure 3. 2D gel electrophoresis (pH 3−10) of the E. coli proteome
stained with (A) Deep Purple Total Protein Stain and (B) compound
4b. Identical concentrations of dye were used in the two gels (1 μg/
mL).
F
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. Details of the 2D gels (triplicates) comparing the
background peak intensities (arrows), slopes, and surface textures
for (A) 4b and (B) Sypro Ruby.
for Sypro Ruby). However, there were some noteworthy
differences in staining that were further investigated by mass
spectrometry. We found that 10 spots (eight proteins) were
stained more intensely with Sypro Ruby but 21 spots (16
proteins) were stained more intensely with 4b (Figure S12 in
the Supporting Information). Identification of these differ-
entially stained proteins showed that neither the aliphatic index
nor the hydropathy value explained the differential labeling
(Table 4). The contents of amino acids were then compared.
Figure 5. Photobleaching studies of Deep Purple (containing 1) and
epicocconone analogues 4a−d.
Information), indicating that the underivatized dye in the
background of the gel is bleached faster by UV light than the
stain colocated with protein. This may be due to the fact that
4b has significant fluorescence even in the absence of protein
(Figure 1A), leading initially to a high background (where there
is no protein) in gels stained with this analogue. Meanwhile, the
enamine of 4b formed in the presence of protein is more stable
toward photobleaching than free 4b, leading to an increase in
S/N over time.
Table 4. Statistical Analysis of Differential Staining Between
4b and Sypro Ruby (SR)
mean SE
4b
SR
p value
aliphatic index
hydropathy
92.6 1.7
−0.226 0.061
0.077 0.005
0.047 0.006
0.012 0.002
90.2 1.8
−0.143 0.056
0.047 0.007
0.063 0.010
0.030 0.005
0.393
0.384
0.001
0.145
0.0006
CONCLUSIONS
■
percentage Lys
percentage Arg
percentage His
In conclusion, we have taken the first approaches to the total
synthesis of the natural product epicocconone (1) by preparing,
in five steps, several analogues that possess distinct advantages
over the natural product in terms of quantum yield and
stability. The synthesis design is general and straightforward,
providing facile access to libraries of polyfunctional fluoro-
phores with large Stokes shifts based on a lead from nature.
The resulting epicocconone analogues, especially 4b, were
found to be excellent latent fluorophores that are useful in
protein detection in 1D and 2D gel electrophoresis with
superior sensitivity and stability compared with the leading
commercial products Sypro Ruby and Deep Purple, respec-
tively. These enhanced detection properties are expected to
enable new protein discovery opportunities. For example, a gel
stain better than Deep Purple with a biotin tag will open a new
dimension of practical applications to visualize proteins in gels
and on membranes, beads, or protein arrays. Such biotin affinity
tags will also enable the isolatation, purification, manipulation,
processing, and analysis of captured proteins that otherwise
would escape detection.
As might be expected, the percentage of lysine was higher for
protein spots specifically stained with 4b (p = 0.001). In
contrast, Sypro Ruby was more sensitive to histidine content (p
= 0.0006), a feature that has not been reported previously. This
underlines the fact that epicocconone analogues more heavily
label proteins that are rich in lysines and that Sypro Ruby (a
ruthenium complex)4 could be expected to complex histidine-
rich proteins.
As Deep Purple has been noted to suffer from photo-
bleaching when used to manually pick protein spots from gels
using a transilluminator, we were keen to assess the
performance of the analogues to see whether replacement of
the triene improved their photostability. Repeating the work of
Smejkal,11 we found Deep Purple, which contains epicocco-
none (1), to have a half-life of 11 min over 60 W UV irradiation
as previously reported. The influence of substitution on the
photostability is clearly shown by the observations that
compound 4a (with an electron-rich p-anisyl substituent) was
photobleached faster than 1 whereas electron-poor 4b
(naphthyl group, t1/2 = 33 min) and 4c (2-chloropyridyl
group, t1/2 = 45 min) were substantially more stable than 1
(Figure 5). These results clearly show the influence of the side
chain on the photostability even though the side chain is not
directly part of the emissive S1 state (Figure 1). An interesting
observation was that for 4b the signal-to-noise ratio (S/N)
actually increased with time (Figure S9 in the Supporting
EXPERIMENTAL SECTION
Synthesis. See the Supporting Information
■
DFT Calculations. DFT calculations were run using Turbomole
6.6 (Cosmologic GmbH & Co). Theoretical ground-state geometries
for all conformations were calculated at the DFT-D3//BP86/TZVPP
level19c,d,25 using a continuum solvent model (COSMO)20 for toluene.
Frequency analysis confirmed that all structures were stationary points.
1D Gel Electrophoresis. The 12% Bis-Tris Novex NuPage
(Invitrogen, NP0342BOX) 1 mm thick gels were loaded with the
dilution series (5 μL) of GE low-molecular-weight markers (GE
G
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Healthcare, 17-0446-01) to obtain the concentrations indicated in
Table S1 in the Supporting Information. The gels were run in Xcell
SureLock minigel systems (Invitrogen, EI0001) at 150 V for
approximately 65 min (buffer front just off gel) using MES buffer
(50 mM MES, 50 mM Tris, 1 mM EDTA, 0.1% SDS, pH 7.3). The
gels were then removed and fixed in 15% (v/v) ethanol with 1% (w/v)
citric acid (100 mL) on a rocker for 1 h and then overnight with fresh
fixative. The next day, gels were placed into fresh fixative (100 mL) for
1 h, drained, and then stained in 100 mM sodium borate buffer (50
mL; pH 10.9) containing either Deep Purple Total Protein Stain (GE
Healthcare, RPN6306; 250 μL) or one of the analogues (50 μL of a 1
mg/mL solution in DMSO). These concentrations resulted in
approximately 1 μg/mL active fluorophore in each case. After 1 h of
staining, the gels were destained in 15% ethanol (100 mL) for 30 min
and transferred back into fresh fixative [100 mL; 15% (v/v) ethanol,
1% (w/v) citric acid] for 30 min. All of the gels were imaged using a
Typhoon Trio imager (GE, 63-0055-87) with a 532 nm Nd:YAG laser,
540 PMT, 610BP30 filter, 100 μm resolution, and normal sensitivity.
The Typhoon scans (.gel files) were analyzed using IMageQuantTL
(v7, GE Healthcare), and band volumes and background were
calculated using the 1D gel band analysis package (Figures S5 and S6
in the Supporting Information).
Photobleaching. 1D gels (as above) were subjected to UV light
by placing them onto a standard transilluminator (4 × 15 W; 312 nm
tubes). The gels were imaged after each interval and then placed back
into cold fixative (100 mL) for 3 min to allow them to cool. Each gel
was exposed to UV light for 60, 146, 240, 330, 630, and 990 s
(cumulative exposure). The Typhoon scans (.gel files; Figures S7 and
S8 in the Supporting Information) were analyzed using IMage-
QuantTL (v7, GE healthcare), and band volumes and background
(between bands) were calculated using IMageQuantTL. The average
volume of each protein band over background was fitted to a one-
phase exponential decay (Graphpad Prism v5.0), and the pseudo-first-
order decays rate was calculated.
2D Gel Electrophoresis. Four 24 cm pI 3−10 Immobiline dry
strips (GE Healthcare, 17-1234-01) were rehydrated into IEF buffer
(40 mM Tris-HCl, 2 M thiourea, 5 M urea, 2% CHAPS, 2%
sulfobetaine 3−10, 0.002% bromophenol blue; 180 μL per IPG strip)
containing whole-cell lysate from Escherichia coli (100 μg per IPG
strip) for 6 h by laying the strips gel down in a rehydration tray. The
strips were then focused for 48 000 V h using an IPGphor 3 power
pack (GE Healthcare, 11-0033-64) before being stored at −80 °C. The
strips were equilibrated using equilibration buffer (GelCompany,
1019-20; 2 × 10 mL) for 2 × 15 min; the first equilibration also
contained 1% (w/v) DTT, while the second contained 5% (w/v)
iodoacetamide. Separation in the second dimension was performed
with a flatbed Tower electrophoresis unit (Serva, HPE-TS1.01) using
the standard manufacturer’s protocol. 2DGel flatbed 12.5% acrylamide
gels (250 mm × 200 mm × 0.65 mm) from GelCompany (1019-02)
were used following the standard protocol. Briefly, focused IPG strips
were removed from the IAA solution and placed gel side down onto
the stacking gel portion of the 2D gels. Proteins were transferred to
the 2D gel (100 V/28 mA/30 min, 200 V/52 mA/30 min, 300 V/80
mA/10 min), and then the IPG strip was removed. The 2D gel was
run at 15 °C (1500 V/160 mA) for 3.5 h. The gels were then removed,
fixed in 15% (v/v) ethanol, 1% (w/v) citric acid (500 mL) on a rocker
for 90 min, and then stained in 100 mM sodium borate buffer (350
mL; pH 10.9) with either Deep Purple Total Protein Stain (GE
Healthcare, RPN6306; gel A 1.00 mL, gel B 1.75 mL) or analogue 4b
(gel C 350 μL; gel D 100 μL of a 1 mg/mL solution in DMSO). Gels
B and C (Figure S10 in the Supporting Information) both had a final
dye concentration of ∼1 μg/mL in the staining solution. After 1 h of
staining, the gels were destained in 15% (v/v) ethanol (500 mL) for 45
min and transferred to fixative [500 mL; 15% (v/v) ethanol, 1% (w/v)
citric acid] for 45 min. All of the gels were imaged using a Typhoon
Trio imager (GE, 63-0055-87) with a 532 nm Nd:YAG laser, 540
PMT, 610BP30 filter, 100 μm resolution, and normal sensitivity.
Sypro Ruby 2D gels were run on a Protean II xi cell (Bio-Rad) using
a 12.5% (w/v) polyacrylamide resolving gel (width, 16 cm; length, 20
cm; thickness, 0.75 mm). Triplicate gels were fixed overnight, and
proteins were stained using Sypro Ruby according to the
manufacturer’s protocol or as above for compound 4b (Figure S12
in the Supporting Information). Gels were scanned using a ProXpress
fluorimager (PerkinElmer) and analyzed using SameSpots 4.0 software
(Nonlinear Dynamics), and all triplicate images were aligned before
analysis. Gels were stored in water at 4 °C before spot excision.
Protein Identification. After excision of protein spots from 2D
gels, the fragments were washed with water and dried in a SpeedVac
centrifuge for a few minutes. Trypsin digestion (with 10 μL of a 10 ng/
μL trypsin solution, Promega) was performed overnight (PerkinElmer,
MultiPROBE II). The peptides were extracted with H2O/CH3CN
(1:1) for 2 × 15 min, dried, and solubilized in 3% CH3CN/0.1%
HCOOH/water. The peptides were concentrated and separated by
reversed-phase HPLC with a precolumn/analytical column nanoflow
setup (Agilent, HPLC-Chip cube). The peptides were analyzed by
LC−MS and further fragmented after a full survey scan (m/z 300−
2000) using an online QTOF mass spectrometer (Agilent, 6520). MS/
MS experiments on the five most abundant precursor ions were
performed, and the fragmentation data were exported using the
MassHunter qualitative analysis module (Agilent, version B.02.00).
For protein identification, extracted MS/MS peak lists were
compared to the MSDB database restricted to the E. coli ORF protein
database (31797) using the MASCOT Daemon (version 2.1.3) search
engine. All searches were performed with no fixed modification and
with variable modifications for carbamidomethylation of cysteines and
for oxidation of methionines, with a maximum of one missed cleavage.
MS2 spectra were searched with a mass tolerance of 5 ppm for
precursor ions and 0.02 Da for fragment ions, respectively. The protein
identification was validated if two peptides exhibited fragmentation
profile scores higher than the average default value for significance
using MASCOT. For sequence analysis, ProtParam (http://web.
expasy.org/protparam/) was used to calculate the amino acid
composition, aliphatic index, and grand average of hydropathicity
(GRAVY). The aliphatic index of a protein is defined as the relative
volume occupied by aliphatic side chains (alanine, valine, isoleucine,
and leucine). The GRAVY value for a protein was calculated as the
sum of the hydropathy values of all the amino acids divided by the
number of residues in the sequence.
ASSOCIATED CONTENT
■
S
* Supporting Information
Synthetic procedures, NMR spectra, fluorescence data, gel
images, and analyses of gel images. This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
peter.karuso@mq.edu.au
xavier.franck@insa-rouen.fr
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We gratefully acknowledge the Reg
financial support to P.A.P. and A.B and the ANR-BLAN-732-01
grant for financial support and the Australian Research Council
́
ion Haute Normandie for
for support of P.K. We thank the Minister
̀
e des Affaires
Etrangeres (France) and the Department of Innovation,
̀
Industry, Science and Research (Australia) for FAST Grants.
This research was facilitated by access to the Australian
Proteome Analysis Facility established under the Australian
Government’s Major National Research Facilities Program. The
authors thank Kent Taylor (Gratuk Technologies) for running
the 2D gels. Dr. Fei Liu (Macquarie University) is acknowl-
edged for helpful discussions.
H
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(22) Chatterjee, S.; Karuso, P.; Boulange,
́
A.; Peixoto, P. A.; Franck,
REFERENCES
■
X.; Datta, A. J. Phys. Chem. B 2013, 117, 14951−14959.
(23) Deep Purple is a commercial product from GE Healthcare
consisting of ∼1.8% epicocconone specially formulated for 1D and 2D
protein gel staining. The stain has high sensitivity when scanned with
either a 488 or 532 nm laser scanner. See: http://www.fluorotechnics.
com.au/pdf/dige_compat.pdf (accessed July 11, 2014).
(24) Berggren, K.; Chernokalskaya, E.; Steinberg, T. H.; Kemper, C.;
Lopez, M. F.; Diwu, Z.; Haugland, R. P.; Patton, W. F. Electrophoresis
2000, 21, 2509−2521.
(1) Udenfriend, S.; Stein, S.; Bohlen, P.; Dairman, W.; Leimgruber,
̈
W.; Weigele, M. Science 1972, 178, 871−872.
(2) Chandran, S. S.; Dickson, K. A.; Raines, R. T. J. Am. Chem. Soc.
2005, 127, 1652−1653.
(3) Girouard, S.; Houle, M. H.; Grandbois, A.; Keillor, J. W.;
Michnick, S. W. J. Am. Chem. Soc. 2005, 127, 559−566.
(4) Ball, M. S.; Karuso, P. J. Proteome Res. 2007, 6, 4313−4320.
(5) Bell, P. J. L.; Karuso, P. J. Am. Chem. Soc. 2003, 125, 9304−9305.
(6) (a) Moritz, C. P.; Marz, S. X.; Reiss, R.; Schulenborg, T.; Friauf,
E. Proteomics 2014, 14, 162−168. (b) Tang, W. Methods Mol. Biol.
2012, 876, 67−82. (c) Liu, H.; Crooks, R. M. J. Am. Chem. Soc. 2011,
133, 17564−17566. (d) Steinberg, T. H. Methods Enzymol. 2009, 463,
541−563. (e) Karp, N. A.; Feret, R.; Rubtsov, D. V.; Lilley, K. S.
Proteomics 2008, 8, 948−960. (f) Karuso, P.; Crawford, A. S.; Veal, D.
A.; Scott, G. B. I.; Choi, H.-Y. J. Proteome Res. 2008, 7, 361−366.
(g) Choi, H. Y.; Veal, D. A.; Karuso, P. J. Fluoresc. 2006, 16, 475−482.
(h) Mackintosh, J. A.; Veal, D. A.; Karuso, P. Proteomics 2005, 5,
4673−4677. (i) Mackintosh, J. A.; Choi, H.-Y.; Bae, S.-H.; Veal, D. A.;
Bell, P. J.; Ferrari, B. C.; Van Dyk, D. D.; Verrills, N. M.; Paik, Y.-K.;
Karuso, P. Proteomics 2003, 3, 2273−2288.
(25) Bauernschmitt, R.; Haser, M.; Treutler, O.; Ahlrichs, R. Chem.
Phys. Lett. 1997, 264, 573−578.
(7) Coghlan, D. R.; Mackintosh, J. A.; Karuso, P. Org. Lett. 2005, 7,
2401−2404.
(8) (a) Chatterjee, S.; Burai, T. N.; Karuso, P.; Datta, A. J. Phys.
Chem. A 2011, 115, 10154−10158. (b) Panda, D.; Datta, A. J. Chem.
Sci. 2007, 119, 99−104.
(9) Haugland, R. P. Handbook of Fluorescent Probes and Research
Products, 9th ed.; Molecular Probes: Eugene, OR, 2002.
(10) Molecular Probes HandbookA Guide to Fluorescent Probes and
Labeling Technologies, 11th ed.; Life Technologies: Eugene, OR, 2010.
(11) Smejkal, G. B.; Robinson, M. H.; Lazarev, A. Electrophoresis
2004, 25, 2511−2519.
(12) Garavelli, M.; Celani, P.; Yamamoto, N.; Bernardi, F.; Robb, M.
A.; Olivucci, M. J. Am. Chem. Soc. 1996, 118, 11656−11657.
(13) Hirashima, S.-i.; Itoh, A. Green Chem. 2007, 9, 318−320.
(14) (a) Millioni, R.; Tolin, S.; Sbrignadello, S.; Fadini, G.; Puricelli,
L.; Tessari, P.; Arrigoni, G. Nutr., Metab. Cardiovasc. Dis. 2013, 23,
S54−S55. (b) Fey, S. J.; Larsen, P. M. Curr. Opin. Chem. Biol. 2001, 5,
26−33. (c) Wang, H.; Chang-Wong, T.; Tang, H.-Y.; Speicher, D. W.
J. Proteome Res. 2010, 9, 1032−1040. (d) Beseme, O.; Fertin, M.;
Drobecq, H.; Amouyel, P.; Pinet, F. Electrophoresis 2010, 31, 2697−
2704.
(15) (a) Roche, S. P.; Porco, J. A., Jr. Angew. Chem., Int. Ed. 2011, 50,
4068−4093. (b) Van De Water, R. W.; Hoarau, C.; Pettus, T. R.
Tetrahedron Lett. 2003, 44, 5109−5113.
(16) (a) Boulange,
́
A.; Peixoto, P. A.; Franck, X. Chem.Eur. J. 2011,
17, 10241−10245. (b) Franck, X.; Peixoto, P.; Karuso, P. H.
(Universite de Rouen, France; Macquarie University, Australia).
Process for Synthesizing Epicocconone Analogues as Fluorescent
Dyes. WO2011051225A1, 2011. (c) Lebrasseur, N.; Gagnepain, J.;
Ozanne-Beaudenon, A.; Leger, J. M.; Quideau, S. J. Org. Chem. 2007,
72, 6280−6283.
(17) Reber, K. P.; Tilley, S. D.; Sorensen, E. J. Chem. Soc. Rev. 2009,
38, 3022−3034.
(18) (a) Carroll, M. F.; Bader, A. R. J. Am. Chem. Soc. 1953, 75,
5400−5402. (b) Peixoto, P. A.; Boulange,
́
A.; Leleu, S.; Franck, X. Eur.
J. Org. Chem. 2013, 3316−3327.
(19) (a) Treutler, O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346−354.
(b) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor. Chem.
Acc. 1997, 97, 119−124. (c) Weigend, F.; Ahlrichs, R. Phys. Chem.
Chem. Phys. 2005, 7, 3297−3305. (d) Grimme, S.; Antony, J.; Ehrlich,
S.; Krieg, H. J. Chem. Phys. 2010, 132, No. 154104.
(20) Klamt, A.; Schuurmann, G. J. Chem. Soc., Perkin Trans. 2 1993,
̈
̈
799−805.
(21) (a) Syzgantseva, O. A.; Tognetti, V.; Boulange,
A.; Leleu, S.; Franck, X.; Joubert, L. J. Phys. Chem. A 2014, 118, 757−
764. (b) Syzgantseva, O. A.; Tognetti, V.; Joubert, L.; Boulange, A.;
́
A.; Peixoto, P.
́
Peixoto, P. A.; Leleu, S.; Franck, X. J. Phys. Chem. A 2012, 116, 8634−
8643.
I
dx.doi.org/10.1021/ja506914p | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX