Small-molecule fluorescent stains enable the imaging of cellular structures without the need for genetic manipulation. Here, we introduce 2,7-diaminobenzopyrylium (DAB) dyes as live-cell mitochondrial stains excited with violet light. This amalgam of the coumarin and rhodamine fluorophore structures yields dyes with high photostability and tunable spectral properties.
Small-molecule fluorescent stains enable the imaging of cellular structures without the need for genetic manipulation. Here, we introduce 2,7-diaminobenzopyrylium (DAB) dyes as live-cell mitochondrial stains excited with violet light. This amalgam of the coumarin and rhodamine fluorophore structures yields dyes with high photostability and tunable spectral properties.
Fluorescence microscopy
is an essential tool to interrogate biological
structure. A key element in any imaging experiment is the labeling
strategy used to localize a fluorophore to the cellular component
of interest.[1−5] In addition to measuring the position and movement of specific biomolecules,
cellular imaging experiments often involve the visualization of different
organelles to quantify their dynamics[6] or
provide useful subcellular reference marks.[7−9] Fusing a fluorescent
protein to a targeting motif can allow labeling of cellular organelles
but requires expression of an exogenous molecule. Fluorescent reagents
with affinities for organelle-specific biomolecules can allow imaging
without genetic manipulation but typically involve preparation of
a small-molecule fluorophore conjugated to an antibody or drug. An
alternative labeling strategy is the use of fluorescent stains that
accumulate in specific organelles due to the different chemical environments
found in these distinct subcellular regions. Examples include tertiary
amine-containing dyes accumulating in acidic lysosomes[10] or hydrophobic fluorophores partitioning into
lipid droplets.[11]A widely used fluorescent
staining strategy involves mitochondria,
whose double membrane structure reflects their role as the powerhouse
of the cell. The proteins that comprise the electron transport chain
reside in the inner mitochondrial membrane that separates the matrix
from the intermembrane space. Their activity drives protons across
the inner membrane, resulting in a large voltage difference between
these two compartments. This unique membrane potential drives the
accumulation of lipophilic cations into the matrix or inner membrane.
This was first observed with Rhodamine 123 (1),[12] where esterification of the standard ortho-carboxyl group found in rhodamines endows the molecule
with a fixed cationic charge. The strategy was expanded to tetramethylrhodamine
methyl ester (TMRM, 2), yielding a red-shifted mitochondrial
stain.[13] This general idea was refined
with the development of MitoTracker Orange (3), in which
the carboxyl ester functionality found in rhodamines 1 and 2 is discarded entirely.[7] Compound 3 also incorporates a chloromethyl moiety
to allow formation of a glutathione adduct, thereby trapping the fluorophore
inside the cell.[14]The majority of
fluorescent mitochondrial stains are based on rhodamine
(e.g., 1–3) and cyanine structures.[7] These dyes exhibit relatively long absorption
maxima (λabs) and fall into the standard blue (488
nm), green-yellow (560 nm), and red (640 nm) excitation windows used
in fluorescence imaging. Mitochondrial stains excited with violet
light (405 nm) have received less attention since there is no general
cationic dye scaffold with an excitation maximum in this wavelength
range. We sought, and now report, a new class of mitochondrial stains
based on 2,7-diaminobenzopyrylium (DAB) dyes.
Results and Discussion
To create a violet-excited mitochondrial stain, we first considered
coumarin dyes, which remain the most utilized fluorophores excited
by ultraviolet (UV) and violet light. The simplest fluorescent coumarins
are 7-hydroxy derivatives such as 4-methylumbelliferone (4, Figure b). We noted
that the classic fluorophore fluorescein (5) is effectively
the phenylogous derivative of 4. Similarly, 7-aminocoumarins
such as Coumarin 1 (6) are structurally consonant with
rhodols like 7. Inspired by the structural relationship
between compounds 4/5 and 6/7, we considered 2,7-diaminobenzopyrylium (DAB) structures
exemplified by the tetraethyl derivative 8; this is the
“coumarin-sized” analog of Rhodamine B (9). Although iminocoumarins have received some attention as dyes[15] and indicators,[16] the cationic 2,7-diaminobenzopyrylium fluorophore scaffold is essentially
unexplored, with a lone report in the Soviet chemistry literature
from 1989.[17]
Figure 1
2,7-Diaminobenzopyrylium
(DAB) dyes as potential mitochondrial
stains. (a) Chemical structures of mitochondrial stains 1–3. (b) Pairs of structurally consonant dyes:
4-methylumbelliferone (4) and fluorescein (5); Coumarin 1 (6) and N,N-diethylrhodol (7); 2,7-diaminobenzopyrylium 8 and Rhodamine B (9).
2,7-Diaminobenzopyrylium
(DAB) dyes as potential mitochondrial
stains. (a) Chemical structures of mitochondrial stains 1–3. (b) Pairs of structurally consonant dyes:
4-methylumbelliferone (4) and fluorescein (5); Coumarin 1 (6) and N,N-diethylrhodol (7); 2,7-diaminobenzopyrylium 8 and Rhodamine B (9).We began our investigation by synthesizing the known tetraethyl
DAB dye 8 starting from Coumarin 1 (6, Scheme a).[17] Treatment with Et3OBF4 generates
a 2-ethoxychromenylium intermediate that reacts with diethylamine
to give 8, which we obtained in 41% yield. Based on the
success of this simple synthetic protocol, we prepared additional
DAB dyes. We transformed the bright azetidinylcoumarin[18]10 into the diazetidinyl dye 11 (Scheme b). We also explored dyes derived from Coumarin 102 (12); reaction with diethylamine, azetidine, or piperidine yielded compounds 13–15 (Scheme c).
Scheme 1
Synthesis of dyes 8 (a), 11 (b), and 13–15 (c)
We then evaluated the spectral properties of
the DAB dyes 8, 11, and 13–15 in phosphate-buffered saline (PBS; Table , Figure a–c) comparing them to the parent coumarin fluorophores 6, 10, and 12. In general, the transformation
of the carbonyl group into an iminium moiety elicits a bathochromic
shift of ∼50 nm in λabs and a shift of 25–35
nm in fluorescence emission maxima (λem). Thus, all
the DAB dyes exhibit λabs > 400 nm, absorbing
in
the violet-blue region of the visible spectrum. The reduced shift
in λem results in smaller Stokes shifts for the DAB
fluorophores relative to the coumarin starting materials. Despite
this decrease, the Stokes shifts of the DAB dyes (65–100 nm)
remain substantially larger than those of fluoresceins or rhodamines
(∼25 nm). Finally, for each matched pair, the DAB congeners
show substantially higher photobleaching time constants (tb) and average number of photons emitted before photobleaching
(Np) relative to the corresponding coumarin
dye.
Table 1
Spectral Properties of DAB Dyesa
All values measured in PBS, pH 7.4,
solution.
Figure 2
Spectra and photobleaching
of representative DAB dyes. (a–c)
Normalized absorption (abs) and fluorescence emission (fl) of (a) 6 and 8, (b) 10 and 11, and (c) 12–15. (d–f) Plot
of normalized fluorescence vs irradiation time of (d) 6 and 8, (e) 10 and 11, and
(f) 12–15 with monoexponential fit.
All values measured in PBS, pH 7.4,
solution.Spectra and photobleaching
of representative DAB dyes. (a–c)
Normalized absorption (abs) and fluorescence emission (fl) of (a) 6 and 8, (b) 10 and 11, and (c) 12–15. (d–f) Plot
of normalized fluorescence vs irradiation time of (d) 6 and 8, (e) 10 and 11, and
(f) 12–15 with monoexponential fit.Comparison of specific dye pairs reveal more nuanced
differences.
The transformation of Coumarin 1 (6) to DAB dye 8 yields a shift of 45 nm in λabs and 26
nm in λem, resulting in a Stokes shift of 65 nm (Table , Figure a). Conversion of coumarin 10 into diazetidinyl DAB derivative 11 results
in a bathochromic shift of 48 nm in λabs and a difference
of 34 nm in λem (Table , Figure b). This larger shift in emission maximum combined
with the properties of the parent coumarin 10 gives a
Stokes shift of 99 nm for 11. For DAB dyes 13–15 derived from Coumarin 102 (12), the λabs does not depend on the different secondary
amine auxochromes (Table , Figure c).
The fluorescence emission does vary with structure; the acyclic N,N-diethylamino derivative 13 exhibits λem = 513 nm, which is 10 nm shorter than 14 and 15, resulting in a smaller Stokes shift
of 68 nm.Across the series, we found that the absorptivity
of the DAB dyes
is substantially higher than the parent coumarin fluorophores, with
1.5–2-fold increases in extinction coefficient at λabs (ε, Table ). The transformation of the coumarin oxygen into an iminium
moiety elicits variable effects on fluorescence quantum yield (Φf). For the relatively dim Coumarin 1 (6; Φf = 0.06) conversion to DAB dye 8 increases quantum
yield (Φf = 0.16). In contrast, diazetidinyl dye 11 shows a modestly lower quantum yield (Φf = 0.79) compared to the bright azetidinyl coumarin 10 (Φf = 0.96). The DAB dyes 13–15 exhibit lower Φf values compared to 12. These modest decreases in Φf for compounds 11 and 13–15 are balanced
by the larger ε values, resulting in higher molecular brightness
(ε × Φf) for the DAB dyes relative to
the corresponding coumarins.As mentioned above, the DAB dyes
exhibit increased photostability
compared to their coumarin congeners. Dyes 8 and 11 gave 1.5–2-fold longer tb values compared to coumarins 6 and 10 (Table , Figure d,e). For compounds 13–15, the photostability improvement is greater,
with these DAB compounds showing consistent tb values that are 5-fold higher than the parent coumarin 12 (Table , Figure d). Since
photobleaching reactions stem from excited states, it is difficult
to compare photobleaching time constants between dyes with different
fluorescence quantum yields and lifetimes; this is reflected in the
different bleaching rates observed across the various dye types (Table , Figure d–f). We therefore calculated
the average total photons (Np) emitted
by each dye.[19] This parameter highlights
the increased photostability of the DAB fluorophores and revealed
remarkably consistent photostability for compounds 11 and 13–15 with an average of ∼8300
photons/dye (Table ).We then evaluated the stability of the iminium linkage in
aqueous
solution under “dark” conditions and under illumination
with violet (405 nm) light. Monitoring with tandem liquid chromatography–mass
spectrometry (LC–MS) revealed that all the DAB dyes (8, 11, and 13–15) show excellent stability with minimal iminium hydrolysis after
48 h at pH 7.4 in the absence of light (Figure a,b, Figure S1). We also evaluated the stability of dye 15 in different
buffer conditions, observing modestly higher rates of hydrolysis at
elevated pH or in cell culture media containing serum (Figure S2). We then undertook a comprehensive
photochemistry study of 15 using 405 nm illumination
and analysis by LC–MS (Figure c,d, Figure S3) The photochemical
reactions of 15 are similar to those observed with parent
coumarin 12 where oxidation appears to be centered on
the julolidine ring system;[20,21] we did not detect any
photochemically driven oxidation on the piperidine ring. Overall,
these data show that the DAB dyes exhibit reasonable chemical stability
and the iminium motif is not susceptible to photochemical degradation.
Figure 3
Stability
of DAB dye 15. (a) Spontaneous hydrolysis
of 15 to form coumarin 12. (b) LC–MS
chromatograms of 15 at t = 0 (top), 15 at t = 48 h (middle), and standard 12 (bottom) in PBS, pH 7.4. (c) Photochemistry of 15 to form oxidized adducts 16–18.
(d) LC–MS chromatograms in PBS, pH 7.4, of 15 in
the absence of light (top), 15 after 5 min illumination
with 405 nm light (middle), and 15 after 20 min illumination
with 405 nm light (bottom).
Stability
of DAB dye 15. (a) Spontaneous hydrolysis
of 15 to form coumarin 12. (b) LC–MS
chromatograms of 15 at t = 0 (top), 15 at t = 48 h (middle), and standard 12 (bottom) in PBS, pH 7.4. (c) Photochemistry of 15 to form oxidized adducts 16–18.
(d) LC–MS chromatograms in PBS, pH 7.4, of 15 in
the absence of light (top), 15 after 5 min illumination
with 405 nm light (middle), and 15 after 20 min illumination
with 405 nm light (bottom).Given the propensity of cationic rhodamines to accumulate in the
mitochondria (Figure a), we investigated the similarly charged DAB dyes as live-cell mitochondrial
stains. We incubated live U2OS cells with dyes 8, 11, and 13–15 at 200 nM and
co-stained the mitochondria either using MitoTracker Deep Red (Figure ) or by transient
transfection of HaloTag–TOMM20 and staining with the far-red
fluorogenic label Janelia Fluor 635-HaloTag ligand[22] (Figure S4). In both experiments,
we observed similar staining patterns between the DAB dyes and the
established far-red stain or genetically encoded label, confirming
our hypothesis that this positively charged fluorophore scaffold would
accumulate in mitochondria. Imaging using the same settings revealed
that the julolidine-containing derivatives 13–15 showed brighter staining, perhaps due to increased lipophilicity
of the compact cationic structure.
Figure 4
DAB dyes stain mitochondria. Chemical
structures of DAB dyes 8, 11, and 13–15 and corresponding fluorescence images
of U2OS cells co-stained with
the DAB dye (200 nM) and MitoTracker Deep Red (100 nM). Scale bar,
10 μm.
DAB dyes stain mitochondria. Chemical
structures of DAB dyes 8, 11, and 13–15 and corresponding fluorescence images
of U2OS cells co-stained with
the DAB dye (200 nM) and MitoTracker Deep Red (100 nM). Scale bar,
10 μm.In these cellular imaging experiments,
we noted that the cellular
intensity of even the best DAB mitochondrial stain 15 rapidly decreased upon media exchange. To further improve this reagent,
we prepared a dimer derivative of this molecule by reacting Coumarin
102 (12) with dipiperidine 19 to yield “diDAB” 20 (Figure a). This design is predicated on two concepts. First, the relatively
long Stokes shift of the parent dye 15 (Table , Figure c) should minimize FRET between the two fluorophore
units and preserve fluorescence quantum yield. Second, the presence
of two cationic moieties per molecule of 20 should improve
mitochondrial retention. Examination of the chemical properties of 20 revealed a slightly higher rate of hydrolysis relative
to 15 in different pH conditions even when considering
the presence of two iminium groups (Figure S5).
Figure 5
Dimerization of DAB dye 15 improves performance as
a mitochondrial stain. (a) Synthesis of diDAB 20. (b)
Normalized absorption (abs) and fluorescence emission (fl) of 15 and 20. (c) Plot of cellular fluorescence
vs time of cells incubated with 15, 20,
or MitoView 405 during photobleaching; error bars indicate SEM. (d)
Live U2OS cells incubated with DAB 15 (200 nM), diDAB 20 (200 nM), or MitoView 405 (100 nM) after 0, 1, or 2 dye-free
media exchange washes. Scale bar, 10 μm.
Dimerization of DAB dye 15 improves performance as
a mitochondrial stain. (a) Synthesis of diDAB 20. (b)
Normalized absorption (abs) and fluorescence emission (fl) of 15 and 20. (c) Plot of cellular fluorescence
vs time of cells incubated with 15, 20,
or MitoView 405 during photobleaching; error bars indicate SEM. (d)
Live U2OS cells incubated with DAB 15 (200 nM), diDAB 20 (200 nM), or MitoView 405 (100 nM) after 0, 1, or 2 dye-free
media exchange washes. Scale bar, 10 μm.We then measured the spectral properties of the dye dimer. Compound 20 exhibited similar absorption (λabs = 447
nm) and fluorescence emission (λem = 524 nm) spectra
compared to DAB monomer 15 (Figure b); both dyes showed a linear relationship
between absorption and concentration ≤20 μM (Figure S6). In aqueous solution, the diDAB 20 did not show the expected 2-fold increase in absorptivity
but exhibited ε = 45 200 M–1 cm–1 along with a modestly lower Φf =
0.45 (Table S1). To reconcile this observation
and mimic the nonpolar environment of the mitochondrial inner membrane,
we measured the spectral properties of compounds 15 and 20 in sodium dodecyl sulfate (SDS) micelles and dioxane/water
mixtures (Table S1). We observed the expected
higher ε = 78 900 M–1 cm–1 in PBS containing 0.1% w/v SDS along with a larger Φf = 0.83; compound 20 is also brighter in aqueous dioxane.
Although dye 15 also showed higher absorptivity and fluorescence
quantum yield in these nonpolar conditions relative to PBS, the effect
was less pronounced. We postulate that the relatively low ε
and Φf observed for compound 20 in aqueous
solution is due, in part, to intramolecular interactions between the
two chromophore units. These interactions could be reduced in the
more hydrophobic environment of SDS micelles or dioxane/water mixtures,
resulting in higher absorptivity and fluorescence quantum yield.We then compared the parent DAB 15, diDAB 20, and the commercial violet-excited (and structurally mysterious)
MitoView 405 in live-cell experiments. Although MitoView 405 was modestly
brighter than DAB 15 or diDAB 20 upon initial
application, the dye bleached rapidly in live cells in our hands,
preventing acquisition of a full confocal microscopy stack. The DAB
compounds exhibited substantially higher resistance to photobleaching
with the diDAB 20 showing the best overall photostability
(Figure c). As expected,
all the dyes showed excellent mitochondrial staining upon initial
application (Figure d). Upon media exchange, however, the DAB 15 signal
rapidly decreased whereas the diDAB 20 and MitoView 405
were retained after this cell washing protocol.
Conclusion
In
summary, we demonstrate that the 2,7-diaminobenzopyrylium (DAB)
framework is a modular scaffold for the synthesis of mitochondrial
stains excited with violet light. These atom-efficient imaging reagents
can be prepared from the broad palette of 7-aminocoumarin dyes with
different N-substitution patterns (Scheme ). Although the spectral properties can be
tuned by choosing different coumarin starting materials, the structure
of the secondary amine reactant has only minor effects on the properties
of the resulting DAB dyes: use of azetidine or piperidine gave fluorophores
with similar quantum yields and photostability (Table ). This is in contrast to rhodamine dyes,
where azetidine or piperidine auxochromes can elicit 10-fold changes
in Φf.[18] The DAB dyes
exhibit higher absorptivity and photostability than their coumarin
parent dyes (Table , Figure ), show
reasonable chemical stability (Figure ), and are effective mitochondrial stains (Figure ); dimerization of
the DAB dye affords a stain with better cellular retention (Figure ). Looking forward,
the utility of these bright, photostable, and biocompatible “mini-rhodamines”
(Figure ) could be
expanded beyond mitochondrial stains. The stability of the iminium
moiety could be improved through intramolecular cyclization to develop
new conjugatable fluorescent labels or photolabile groups. The slow
rate of hydrolysis could also be tuned and exploited to release coumarin-based
drugs.[23] Overall, the DAB dyes represent
an underutilized chemical scaffold worthy of further attention.
Authors: Ahmed S Abdelfattah; Takashi Kawashima; Amrita Singh; Ondrej Novak; Hui Liu; Yichun Shuai; Yi-Chieh Huang; Luke Campagnola; Stephanie C Seeman; Jianing Yu; Jihong Zheng; Jonathan B Grimm; Ronak Patel; Johannes Friedrich; Brett D Mensh; Liam Paninski; John J Macklin; Gabe J Murphy; Kaspar Podgorski; Bei-Jung Lin; Tsai-Wen Chen; Glenn C Turner; Zhe Liu; Minoru Koyama; Karel Svoboda; Misha B Ahrens; Luke D Lavis; Eric R Schreiter Journal: Science Date: 2019-08-01 Impact factor: 47.728
Authors: Jonathan B Grimm; Brian P English; Jiji Chen; Joel P Slaughter; Zhengjian Zhang; Andrey Revyakin; Ronak Patel; John J Macklin; Davide Normanno; Robert H Singer; Timothée Lionnet; Luke D Lavis Journal: Nat Methods Date: 2015-01-19 Impact factor: 28.547
Authors: Roman S Erdmann; Stephanie Wood Baguley; Jennifer H Richens; Rebecca F Wissner; Zhiqun Xi; Edward S Allgeyer; Sheng Zhong; Alexander D Thompson; Nicholas Lowe; Richard Butler; Joerg Bewersdorf; James E Rothman; Daniel St Johnston; Alanna Schepartz; Derek Toomre Journal: Cell Chem Biol Date: 2019-02-07 Impact factor: 8.116
Authors: Jonathan B Grimm; Anand K Muthusamy; Yajie Liang; Timothy A Brown; William C Lemon; Ronak Patel; Rongwen Lu; John J Macklin; Philipp J Keller; Na Ji; Luke D Lavis Journal: Nat Methods Date: 2017-09-04 Impact factor: 28.547