A side-chain engineering strategy for constructing fluorescent dyes with direct and ultrafast self-delivery to living cells

Article abstract of DOI:10.1039/c9sc05875c

Organic fluorescent dyes with excellent self-delivery to living cells are always difficult to find due to the limitation of the plasma membrane having rigorous selectivity. Herein, in order to improve the permeability of dyes, we utilize a side-chain engineering strategy (SCES): adjusting the side-chain length of dyes to fine-tune the adsorption and desorption processes on the membrane-aqueous phase interfaces of the outer and inner leaflets of the plasma membrane. For this, a family of fluorescent derivatives (SPs) was prepared by functionalizing a styryl-pyridinium fluorophore with alkyl side-chains containing a different carbon number from 1 to 22. Systematic experimental investigations and simulated calculations demonstrate that the self-delivery rate of SPs with a suitable length side-chain is about 22-fold higher in SiHa cells and 76-fold higher in mesenchymal stem cells than that of unmodified SP-1, enabling cell-imaging at an ultralow loading concentration of 1 nM and deep penetration in turbid tissue and in vivo. Moreover, the SCES can even endow a membrane-impermeable fluorescent scaffold with good permeability. Further, quantitative research on the relationship between Clog?P and cell permeability shows that when Clog?P is in the range of 1.3-2.5, dyes possess optimal permeability. Therefore, this work not only systematically reports the effect of side-chain length on dye delivery for the first time, but also provides some ideal fluorescent probes. At the same time, it gives a suitable Clog?P range for efficient cellular delivery, which can serve as a guide for designing cell-permeant dyes. In a word, all the results reveal that the SCES is an effective strategy to dramatically improve dye permeability.

Full text of DOI:10.1039/c9sc05875c

Chemical
Science
View Article Online
View Journal
EDGE ARTICLE
A side-chain engineering strategy for constructing
fluorescent dyes with direct and ultrafast self-
delivery to living cells†
Cite this: DOI: 10.1039/c9sc05875c
All publication charges for this article
have been paid for by the Royal Society
of Chemistry
Lifang Guo,a Chuanya Li,a Hai Shang,b Ruoyao Zhang,a Xuechen Li,a Qing Lu,a
a
c
ad
Xiao Cheng,c Zhiqiang Liu,
Jing Zhi Sun
and Xiaoqiang Yu
*
*
*
Organic fluorescent dyes with excellent self-delivery to living cells are always difficult to find due to the
limitation of the plasma membrane having rigorous selectivity. Herein, in order to improve the
permeability of dyes, we utilize a side-chain engineering strategy (SCES): adjusting the side-chain length
of dyes to fine-tune the adsorption and desorption processes on the membrane–aqueous phase
interfaces of the outer and inner leaflets of the plasma membrane. For this, a family of fluorescent
derivatives (SPs) was prepared by functionalizing a styryl-pyridinium fluorophore with alkyl side-chains
containing a different carbon number from 1 to 22. Systematic experimental investigations and simulated
calculations demonstrate that the self-delivery rate of SPs with a suitable length side-chain is about 22-
fold higher in SiHa cells and 76-fold higher in mesenchymal stem cells than that of unmodified SP-1,
enabling cell-imaging at an ultralow loading concentration of 1 nM and deep penetration in turbid tissue
and in vivo. Moreover, the SCES can even endow a membrane-impermeable fluorescent scaffold with
good permeability. Further, quantitative research on the relationship between Clog P and cell permeability
shows that when Clog P is in the range of 1.3–2.5, dyes possess optimal permeability. Therefore, this work
not only systematically reports the effect of side-chain length on dye delivery for the first time, but also
provides some ideal fluorescent probes. At the same time, it gives a suitable Clog P range for efficient
cellular delivery, which can serve as a guide for designing cell-permeant dyes. In a word, all the results
reveal that the SCES is an effective strategy to dramatically improve dye permeability.
Received 19th November 2019
Accepted 26th November 2019
DOI: 10.1039/c9sc05875c
rsc.li/chemical-science
and vehicular (liposome fusion).8–10 Unfortunately, these avail-
able strategies require mandatory treatment of cells, resulting
Introduction
in irreversible destruction of living cells. Then, Tsien et al.
developed a noninvasive method to deliver organic dyes into
living cells, via the AM esterication of carboxyl groups.11 In
comparison, this strategy avoids direct damage to live cells and
the operation procedure is facile. However, it is only applicable
to modifying dyes with carboxyl or hydroxyl groups, and
partially esteried compounds will release highly toxic formal-
dehyde accompanied by hydrolysis within cells.12,13 Recently,
the transmembrane peptide method has also been used to
transport exogenous substances into the cells. However, the
process is affected by not only the peptide sequence and phys-
icochemical properties, but also the local peptide concentra-
tion, local lipid composition, etc.14 Therefore, a simple,
universal, and harmless dye delivery strategy avoiding the above
problems is very necessary.
Organic uorescent dyes have become indispensable tools in
biology, pathology, medicine, and diagnosis, in the quest to
investigate signicant bioevents and visualize ions, active
molecules, biomacromolecules, and organelles in living cells
under uorescence microscopy.1–5 However, due to the rigorous
selectivity of the plasma membrane to substances entering the
cells, the self-delivery of many extraneous organic dyes to living
cells is oen very poor.6,7 To this end, various strategies have
been explored, including mechanical (microinjection), chem-
ical (ATP, Triton X-100, and saponin), electrical (electric shock),
aCenter of Bio & Micro/Nano Functional Materials, State Key Laboratory of Crystal
Materials, Shandong University, Jinan 250100, P. R. China. E-mail: yuxq@sdu.edu.
cn; zqliu@sdu.edu.cn
bInstitute of Robotics, Shanghai Jiao Tong University, Shanghai 200240, P. R. China
cMoE Key Laboratory of Macromolecule Synthesis and Functionalization, Department
of Polymer Science and Engineering, Zhejiang University, Hangzhou 310027, P. R.
China. E-mail: sunjz@zju.edu.cn
Generally, foreign organic dyes pass through the plasma
membrane via a simple diffusion mechanism. The trans-
membrane process can be divided into three steps:15–17 (1)
adsorption to the membrane–aqueous phase interface of the
outer leaet of the plasma membrane; (2) passing through the
lipid bilayer to the opposite interface; (3) desorption from this
dAdvanced Medical Research Institute, Shandong University, Jinan 250100, P. R. China
† Electronic supplementary information (ESI) available. CCDC 1904261–1904263.
For ESI and crystallographic data in CIF or other electronic format see DOI:
10.1039/c9sc05875c
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
interface into cells. Amongst these, the second step is consid- assumptions should be rational, many important research
ered rapid and easy, while the rst and third steps are the rate- studies and experimental data are still lacking for uorescent
determining steps. As a result, the adsorption onto the outer dyes: (1) the effect of the SCES on the delivery of a uorescent
leaet of the plasma membrane and the subsequent release scaffold and relevant simulated calculations have still not been
from the inner leaet to the cytoplasm determines the delivery systematically investigated; (2) the degree to which the dye's
rate of dye molecules. And the rates of the two processes raise permeability can be improved by the side-chain engineering
different requirements for molecular design. Consequently, fast strategy is still unclear; (3) in particular, can the SCES endow
dye self-delivery should be achieved by tuning the two contra-
dictory requirements.
a
membrane-impermeable uorescent dye with good
permeability?
How can we adjust the rate of adsorption and desorption?
In order to fully address these concerns, herein, we prepare
We know that in drug design, a common method for improving a family of styryl-pyridinium salts (abbreviated as SPs, Scheme
drug permeability is to change its side-chain length.18 In addi- 2) with side chains containing different carbon numbers (CN)
tion, we also found that: (1) for many dyes with a short side- from 1 to 22. Systematic experimental studies and discussion of
chain it is difficult or even impossible to enter living cells, the mechanism reveal that with the increase of the side-chain
such as various reported biosensors19–21 as well as commercial length, the permeability of SPs rst increases and then
Ethidium Bromide (EB) and Propidium Iodide (PI); (2) decreases, which well supports the rational guesses shown in
numerous dyes with a long side-chain are oen stranded in the Scheme 1. Notably, among them, SP-6, SP-8, and SP-10 with
plasma membrane, as evidenced by various plasma membrane appropriate side chains possess ultrafast cellular permeability,
probes.22–24 According to these facts mentioned above, one can
rationally envision that tuning the side-chain length can change
the cell permeability of dyes by inuencing the transmembrane
process. And the expected results are shown in Scheme 1: for
short chain dye molecules, only a small portion can enter the
cell due to poor adsorbability (Scheme 1a); as the side-chain is
increased to a suitable length, dyes have the best permeability
due to the obviously enhanced adsorptive and dissociative
performances (Scheme 1b); upon continuing to extend the side-
chain length, the molecular permeability will greatly decrease
(Scheme 1c) and even become zero (Scheme 1d), owing to the
poor dissociative ability; nally, an overlong chain causes
molecules to be impossible to be adsorbed onto the membrane
surface (Scheme 1e), because the overlong chain makes the dye
Scheme 2 The chemical structures (a) and synthetic route (b) of
water-insoluble in an aqueous culture medium. Although these fluorescent dyes (SPs).
Scheme 1 Schematic diagram illustrating the self-delivery behaviors of synthetic molecules with side chains of different lengths (a–e). Here, the
molecular arrangements do not represent their real states and only represent the number of molecules.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
which enables cell-imaging with an unprecedentedly ultralow addition, SP-1 also shows suitable two-photon properties.
concentration of 1 nM and deep penetration in turbid tissues Therefore, a family of SPs from SP-1 to SP-22 was prepared, and
and in vivo. Moreover, the SCES also makes a membrane- the detailed synthesis and structural characterization were
impermeable dye (9E-BMVC1)20 smoothly enter the cells by displayed in the ESI.†
modifying two hexyls. Compared with previously reported
delivery methods, the SCES is simple and straightforward,
avoiding destructive treatment of living cells. Therefore, when
one hopes to deliver a valuable yet low permeability or imper-
meable uorescent scaffold to living cells, one can rst consider
The SCES does not change the chemical and optical properties
and cytotoxicity of the uorescent scaffold
As shown in the 1H NMR spectra of SPs (Fig. 1d and S4†),
introducing a side-chain does not change the chemical shi of
the protons of the conjugated uorophore. Meanwhile, the
single crystal structures of the three compounds (SP-1, SP-2, and
the SCES.
SP-6) also indicate that the different side-chain lengths have
little inuence on the molecular planarity and bond length,
despite the different packing patterns (Fig. 1e and S5–S7†). In
addition, the side-chain also does not change their photo-
Results and discussion
The selection of a representative uorescent scaffold
In order to investigate the effect of the side-chain length on dye
delivery by the commonly used laser scanning confocal
physical characteristics (as summarized in Table
1 and
microscopy (LSCM), there are two requirements that need to be
satised. Firstly, the platform should exhibit a small molecular
structure in order to maximize the effect from sidechains.
Secondly, these dyes should possess appropriate absorption
and emission wavelengths that can be compatible with LSCM.
For these requirements, the classic uorophore styryl-
Fig. S8†). Moreover, all these SPs have an almost uniform molar
extinction coefficient (3), maximum absorption wavelength,
one-photon excitation uorescence (OPEF) and TPEF emission
wavelengths, and even F and d ꢀ F in the same solvent. Also,
their anti-photobleaching abilities are almost the same. Addi-
tionally, different side-chain lengths also do not change the
cytotoxicity of the SPs, as demonstrated by MTT assays in
Fig. S9.† Therefore, the SCES does not change the chemical and
optical properties and cytotoxicity of the uorescent scaffold.
pyridinium
salt
(SP-1),
namely
trans-4-(40-N,N-dia-
lkylaminostyryl)-N-methylpyridinium (abbreviated as SP-1 in
Scheme 2), can be utilized as the dye platform.25–28 On one hand,
SP-1 is easy to synthesize and has a pyridinium moiety that can
be readily modied with side-chains of different lengths by
a simple one-step reaction. On the other hand, as shown in
The effect of the SCES on the delivery behaviours of SP dyes
Fig. 1a–c, Table S1 and Fig. S1–S3,† SP-1 can be excited at To investigate the delivery manner of SPs, we selected SP-1, SP-
473 nm (the commonly used excitation wavelength of LSCM), 6, SP-12, and SP-14 as representatives to incubate with living
and it also exhibits red emission suitable for bioimaging. In SiHa cells at different temperatures (4 ꢁC and 37 ꢁC).6,29 As
Fig. 1 Photographs taken under UV light with 365 nm excitation (a), OPEF (b) and TPEF (c) spectra of SP-1 in various solvents; inset: the enlarged
parts indicated by the red arrow in (b and c); test concentration: 10 mM; lex (OPEF) ¼ 473 nm; lex (TPEF) ¼ 900 nm. (d) Partial 1H NMR spectra of
1
SPs in DMSO-d6 (total H NMR spectra are shown in Fig. S4†). (e) Single-crystal structures of SP-1, SP-2, and SP-6, respectively, with dihedral
angles of 11ꢁ, 16.38ꢁ, and 5.77ꢁ between the two aromatic rings.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
shown in Fig. S10,† the uorescence images indicate that they
can all clearly illuminate the intracellular structures. Moreover,
the intracellular uorescence intensity at 4 ꢁC shows only
ꢁ
a slight decline compared with that at 37 C (Fig. S10b†), indi-
cating that the transport of SPs through the plasma membrane
is not energy-dependent active transport or endocytosis.30 Since
there is no assistance of specic transporters or channel
proteins for transporting SP molecules, SPs do not undergo
facilitated diffusion. Thus, the delivery mechanism for SPs
should be simple diffusion, which is a simple and non-
destructive delivery method to living cells.
The subcellular location of SPs in living cells is speculated to
be around the mitochondria, as cationic dyes tend to accumulate
in the mitochondria due to the large negative mitochondrial
membrane potential (MMP);31 thus MTDR (a commercial mito-
chondrial deep-red probe) was used as a co-stain with SPs. As
shown in Fig. S11,† there are no crosstalk uorescence signals
between SPs and MTDR. From co-localization experiments, we
can see that SPs with CNs from 1 to 14 can all enter living cells
and show over a 0.88 co-localization coefficient with MTDR in
Fig. 2a–h, indicating that they are mitochondria-targeting.
Notably, the time used for staining cells highly depends on the
side-chain length. In particular, SP-6, SP-8, and SP-10 can
perform ultrafast delivery within 30 s, while SP-15 barely enters
the cells and cannot locate an organelle, mainly due to the
obstruction from the long chain (Fig. 2i). SP-16 (Fig. 2j) and SP-18
(Fig. 2k) are stranded in the plasma membrane and cannot enter
the cells, because of the strong interaction between their long
side-chains and the plasma membrane. And SP-22 cannot stain
cells (Fig. 2l), because of the poor water-solubility in the aqueous
culture medium. Therefore, the above results explicitly reveal that
the length of the side-chain has a signicant effect on the delivery
behavior of dyes to living cells.
Assessing the SCES effect on the delivery rate and amount of
SPs by experimental methods
Given that SP-18 stained the plasma membrane and SP-22
aggregated in the aqueous medium, SPs with side-chain CNs in
the range of 1–16 were selected for detailed study of the effect of
side-chain length on dye delivery. Firstly, time-dependent
dynamic analysis (TDDA)32 of SPs was conducted. In this
experiment, living SiHa cells stained with SPs of 200 nM were
immediately observed by LSCM. To avoid overexposed uores-
cence, the imaging parameters such as PMT gain, laser power,
and offset were set to be low and consistent. As shown in Fig. 3a,
aer incubation for 3 min, only SP-6, SP-8, and SP-10 exhibit
visible intracellular uorescence while others have almost no
uorescence, which indicate that more amounts of SP-6, SP-8
and SP-10 enter living cells. When stained for a longer time,
their uorescence can be clearly seen; however, the uorescence
from other SPs is still rather weak. The uorescence intensity at
different times (Fig. 3b) also shows that the three dyes have
obvious increments. In particular, at 40 min, their uorescence
intensity is about 22-fold stronger than that of SP-1. Further,
plots of uorescence intensity versus incubation time (Fig. 3c)
reveal that their intensity growth rates are apparently faster
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
Fig. 2 (a–l) LSCM images of SiHa cells stained with SPs (200 nM) and MTDR (200 nM, 30 min), and their co-localization coefficients (R), as well as
the changes in the staining location with the increase of the side-chain CN; amongst these, (k) and (l) do not show staining of the mitochondria.
Staining time: 10 min for SP-1, SP-2, SP-3, SP-15, and SP-16; 30 s for SP-6, SP-8, and SP-10; 3 min for SP-12 and SP-14; 40 min for SP-18 and SP-
22. lex (SPs) ¼ 473 nm, lem (SPs) ¼ 550–650 nm; lex (MTDR) ¼ 633 nm, lem (MTDR) ¼ 650–750 nm. Bar ¼ 20 mm.
than those of other SPs. In addition, similar results have also by changing the alkyl-chain length and obtain ideal uorescent
been seen in normal MSC: with the increase of the side-chain probes with optimal cell permeability.
length, the uorescence intensity increases rst and then
To obtain more information on the effect of the side-chain
decreases, and when the side-chain length is between 6 and 10, length on dye delivery, the uorescence recovery aer photo-
the intensity reaches the maximum, up to 76-fold higher than bleaching (FRAP) method33 was reasonably adopted.33 Herein,
that of SP-1 (Fig. S12†). Therefore, the above results shed light living cells are well stained with SPs of 200 nM and then the
on an important scientic discovery: we can control dye delivery residual dyes in the medium were washed off with PBS. Next, we
Fig. 3 TDDA for diffusion dynamics of SPs in SiHa cells. (a) LSCM images of cells stained with SPs (200 nM) at different time points (3, 25, and 40
min) and the corresponding DIC images. (b) Time-dependent fluorescence intensity of SPs recorded at different time points. (c) The fluores-
cence intensity plots of SPs over time. lex ¼ 473 nm, lem ¼ 550–650 nm. Bar ¼ 20 mm.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
selected a region of interest to take a picture at their own a relevant physical model can be established for calculating their
appropriate imaging parameter settings (PMT gain, laser delivery rate constants. As shown in Fig. 5a, we assume that SPs
intensity, and offset), as displayed in a1–j1 of Fig. 4A. Then, are substance X, and its extracellular and intracellular concen-
a single cell was enlarged to ll the viewing area (a2–j2 in trations are C1 and C2, respectively. As the extracellular volume is
Fig. 4A). Aer this, the uorescence was rapidly and totally far larger than the intracellular one, the change of C1 can be
extinguished using a high-power laser (a3–j3 in Fig. 4A) and considered negligible and C1 is close to the initial concentration
then imaging parameters were returned to their original levels. C, while C2 will dynamically vary over time. Moreover, the rate
Maintaining the same observation position, new 200 nM SPs constant of substance X from the extracellular aqueous medium
pre-dissolved in the culture medium were added to ensure that to the intracellular space is denoted by k1, and that for the reverse
the concentration of all SPs outside the quenched cell was the process is k2. Because extracellular X appears in the unbound
same at that moment. Over time, newly added dyes will again state while intracellular X is in the bound state in cells due to the
diffuse into the quenched cells and the uorescence signal will denite interactions with intracellular entities, the values of k1
begin to recover. Notably, the control experiments of quenched and k2 should be different. So, the rate equations can be estab-
cells without adding new 200 nM dyes show that the quenched lished, as shown in eqn (1) and (2).
uorescence cannot spontaneously recover during the obser-
dC2ðtÞ
¼ k1C1ðtÞ ꢂ k2C2ðtÞ
C1(t) ¼ C
(1)
(2)
vation time (Fig. 4C), which indicates that restored uorescence
comes from newly added dyes. As shown in a4–j4 and a5–j5 of
Fig. 4A, the uorescence recovery rates of SP-6, SP-8, and SP-10
are much faster relative to those of other SPs. Moreover, from
the uorescence recovery curves in Fig. 4B, the uorescence of
only the three dyes can be restored to 50% relative to the initial
state, demonstrating their excellent membrane permeability.
dt
The nal solution is given in eqn (3).
C2(t) ¼ a(1 ꢂ eꢂbt
)
(3)
(4)
Simulated calculation of the delivery rate constants of SPs
based on an established physical model
where
k1C1
a ¼
;
b ¼ k2
As mentioned above, the delivery manner of SP molecules into
living cells conforms to the simple diffusion mechanism. Thus,
k2
Fig. 4 FRAP analysis for diffusion dynamics of SPs in SiHa cells. (A) (a–j): (1) LSCM images; (2) enlarged images of the white boxes in (1), which
indicate the bleached regions selected; (3) the bleached images; (4 and 5) the representative FRAP images at different times. (B) The fluorescence
intensity ratios of the enlarged cell in the whole FRAP process (0–600 s) relative to the unbleached ones (2). (C) The control experiment in which
the fluorescence was quenched and no new dye was added. lex ¼ 473 nm, lem ¼ 550–650 nm; concentration: 200 nM. Bar (original images) ¼
20 mm; bar (enlarged images) ¼ 10 mm.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
Fig. 5 The calculation of delivery rate constants of SPs in SiHa cells. (a) Top: the established physical model for the transport of SPs through the
membrane; bottom: the corresponding activation energy profile. (b) The experimental intracellular normalized fluorescence intensity curves of
SPs (blue) at different normalized times and the corresponding fitting curves (red). (c) The calculated delivery rate constants of SPs. (d) The
normalized delivery rate constants and FL intensity at 25 min in the TDDA experiment (Fig. 3b) of SPs.
In order to obtain the values of a and b, we adopt the least k2) have been obtained according to eqn (4), as shown in Fig. 5c
squares t model in eqn (5).
and d. Because C1 is a known constant, we can compare k1 by
comparing the values of k1C1. The k1C1 values of SPs with side-
chain lengths of 6, 8, and 10 are dramatically larger than those
of other SPs, up to 23-fold larger than that of SP-1. It is worth
noting that the k2 values of SPs are all very low and close, which
is mainly because once SPs enter the cell, they can rapidly
accumulate in the mitochondria with large a MMP and thus
rarely diffuse out of the cells. As a result, the k1C1 curves can
represent the overall delivery rate changes of SPs, which can
agree well with their experimental normalized FL intensity
variations at 25 min in TDDA (Fig. 3b), as described by the red
solid line and blue dotted line in Fig. 5d. Besides, similar results
also have been obtained in MSC, as shown in Fig. S12b and
S13.† SP-6, SP-8, and SP-10 have 70-fold larger k1C1 values than
SP-1, but almost identical and low k2 values compared with
other SPs. Notably, the k1C1 values in MSC are bigger than those
in SiHa cells, which may be caused by the different cell types.
Therefore, the simulated calculation of delivery rates in
different cells also conrms that dyes with side-chains of suit-
able length can dramatically improve dye delivery.
n
X
2
min f ða; bÞ ¼
½C2ðtiÞ ꢂ Ciꢃ
(5)
i¼0
where C2(ti) is the theoretical intracellular concentration of SPs
at time ti and Ci denotes the experimental intracellular one at
time ti.
Given the low intracellular concentration of SPs, their uo-
rescence intensity should be approximately proportional to the
concentration.34,35 Thus, we can use uorescence intensity (F) to
replace the concentration terms in eqn (5). In order to ensure
the tting accuracy, normalization processing of uorescence
intensity and time terms needs to be carried out. Therefore, we
nally obtain modied eqn (6).
n
X
ꢀ
ꢁ ꢂ
ꢃ
2
F2 t0i ꢂ Fi
(6)
0
min f ða; bÞ ¼
i¼0
where F2ðt0iÞ represents the theoretical normalized intracellular
uorescence intensity of SPs at normalized time t0i and Fi
0
denotes the experimental normalized one at normalized time ti0
from TDDA experiments in Fig. 3c.
According to Fig. 3c, the corresponding normalized curves
can be observed as blue lines in Fig. 5b. Thus, combining these
experimental normalized data and eqn (6), we can obtain the
nal tting curves (red lines in Fig. 5b) as well as parameters
a and b by the gradient descent method.
As shown in Fig. 5b, the nal tting normalized lines (red
lines) can well coincide with the experimental normalized
curves (blue lines). And, the nal values of rate constants (k1 and
The ultrafast delivery of dyes enables ultralow concentration
(1 nM) cell imaging and deep-imaging in tissues and in vivo
Better membrane permeability of dyes means the possibility of
lower loading concentrations. As shown in Fig. 6, ultralow
concentration mitochondrial staining (1 nM) within 20 min has
been successfully realized with SP-6, SP-8, and SP-10, owing to
their ultrafast delivery abilities. However, other SPs as well as
commercial mitochondrial probes (MTG and MTR) cannot do
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Fig. 6 Comparative LSCM images (a), corresponding imaging parameters (b), and relative fluorescence intensity (c) of SiHa cells stained with SP-
6, SP-8, SP-10, and commercial mitochondrial probes (MTG and MTR) of different concentrations (200 nM and 1 nM). Time-dependent fluo-
rescence intensity of SiHa cells stained with MTG (d) or MTR (e) of different concentrations (200 nM, 100 nM, 10 nM, and 1 nM). For SP-6, SP-8,
and SP-10: lex ¼ 473 nm, lem ¼ 550–650 nm. For MTG: lex ¼ 473 nm, lem ¼ 500–600 nm. For MTR: lex ¼ 543 nm, lem ¼ 550–650 nm. Bar ¼ 20
mm.
this, as shown in Fig. S14–S20† and 6. Also, their ultralow also can image living zebrash up to an 86 mm depth (Fig. S24–
concentration staining is not affected by the heterogeneity of S26†). Therefore, these probes with excellent delivery can be
the different cell types (Fig. S21†). Thus, SP-6, SP-8, and SP-10 used to observe targets in tissues and in vivo, thus accelerating
with suitable side chains can serve as ideal probes for ultralow the understanding of intracellular structures in more complex
concentration loading.
and actual biological systems.
The ultrafast delivery is also favorable for easy penetration in
thick biosamples. Given the good TPEF properties of SPs, a two-
photon microscope (TPM) was used to test their penetration
abilities. As shown in Fig. S22,† SPs can clearly image living cells
under a TPM. Moreover, in rat skeletal muscle tissues, they
show a dramatically increased imaging depth (up to 96 mm)
compared to other SPs (Fig. 7a and S23†). Meanwhile, in turbid
tissues stained with SP-6 (Fig. 7b), subtle and regular mito-
chondrial reticulum structures can be recognized unambigu-
ously, and their four different forms36 also can be easily
identied including perivascular mitochondria (PVM), I-band
mitochondria (IBM), ber parallel mitochondria (FPM) and
cross ber connection mitochondria (CFCM). In addition, they
SCES enables conversion of impermeable dyes into permeable
ones
Our group previously reported a carbazole-based double-salt
dye (9E-BMVC1) with two methyls that could not penetrate
living cells.19 Herein, we modify this dye using the SCES and
synthesize a family of 9E-BMVC1 derivatives (9E-BMVCs) as
shown in Fig. 8a. These new dyes show similar optical proper-
ties due to the identical uorophore (Table S2†). By investi-
gating their ability to enter cells and cellular locations (Fig. 8b
and S27†), we found that 9E-BMVC1 cannot enter living cells;
but when the CN reaches 3, the permeability of 9E-BMVC3
increases and its uorescence signal can be detected within
cells; when the CN changes to 6, 9E-BMVC6 rapidly enters the
living cells and stains lamentous structures, which turn out to
be mitochondria by co-localization with MTDR (up to 89% co-
localization coefficient) as shown in Fig. S28;† when the CN
continuously increases to 8 or larger, the corresponding dyes
are blocked by the plasma membrane. Obviously, introducing
a suitable length side-chain can even convert a membrane
impermeable uorescent dye into a permeable one.
In addition, some piecemeal reports also indirectly sup-
ported the applicability of the SCES we have proposed.37–39 For
example, in a previous study, we found that the molecule LAD-1
with methyl always needs a longer time to enter cells in
comparison to the molecule LAD with butyl.36 Moreover,
commercial probes such as TMRM and TMRE with two methyls
Fig. 7 (a) The imaging depth contrast of SPs in skeletal muscle tissues.
(b) TPEF tomography images of mitochondria in rat skeletal muscle
tissues stained with SP-6 (10 mM, 30 min) at different depths ((1–5), 0–
96 mm) and (6) the enlarged image of the yellow box in (3). lex
900 nm, lem ¼ 570–630 nm.
¼
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
cannot enter the cells; when clog P is in the range of ꢂ1.34 to
6.06, the dyes can enter the cell. SP-6, SP-8, and SP-10 with
clog P values of 1.30, 2.36, and 3.42, respectively, show excellent
cell permeability. For 9E-BMVCs with a side-chain on two sides,
when the clog P is lower than ꢂ2.79, the dyes cannot adsorb
onto the membrane surface; in contrast, when clog P is beyond
4.62, the dyes are stuck in the membrane and cannot desorb
from the inner membrane surface. That is, when the clog P is
between ꢂ2.79 and 4.62, 9E-BMVC derivatives can enter the
cells. According to the experimental results, 9E-BMVC6 with
a clog P of 2.50 shows the best cell permeability. To sum up, we
can draw a conclusion that when the clog P is in the range of
1.30–2.50, dyes should possess optimal cell permeability, which
can serve as a guide for designing cell-permeant dyes.
Fig. 8 (a) The molecular structures and synthetic route to 9E-BMVCs.
(b) Top (1–6): LSCM images of SiHa cells stained with 5 mM 9E-BMVCs
for 30 min without washing; bottom: the corresponding magnified
images of the yellow boxes in (1–6). lex ¼ 473 nm, lem ¼ 500–600 nm.
Bar ¼ 20 mm.
Conclusions
In summary, we have demonstrated that a simple SCES can
dramatically improve the delivery of organic dyes. Via an easy
chemical modication, a series of SP dyes with side chains of
different lengths were constructed. Their delivery behaviors
were systematically studied in two different cell lines by cell
imaging experiments and reasonably interpreted by building
a diffusion model. According to the results, the cell delivery rate
can be dramatically changed by simply tuning the side-chain
length. The membrane permeability of SP-6–SP-10 probes
bearing C6–C10 side-chains is above 76-fold higher than that of
SP-1 with a C1 side-chain. Moreover, the cell delivery rate of
uorescent dyes rst increases and then decreases with the
extension of the side-chain length, and clog P values in the
range of 1.30–2.50 should be optimal for the delivery into live
cells. In particular, utilizing the side-chain engineering strategy,
three uorescent probes, SP-6, SP-8, and SP-10 have been con-
structed, which can light up mitochondria in live cells brightly
with a loading concentration of 1 nM. To the best of our
knowledge, this working concentration is much lower than that
of all the reported and commercialized uorescent probes
currently. We expect that the side-chain engineering strategy
can be an efficient guide for designing cell-permeant dyes,
drugs, and other biological reagents.
can cross the plasma membrane more rapidly than rhodamine
123 without an alkyl chain.37
Mechanism discussion on the relationship between the side-
chain length and cellular delivery of dyes
We rst interpret the effect of the side-chain length on cell
permeability from the micro-viewpoint. According to the
molecular transmembrane model, the desorption process is
opposite to the adsorption.15–17 Thus, for efficient cellular
delivery, there is a balance between the two contradictory
requirements. In the adsorption process, the dye molecule
needs to remove the hydration shell (energy barrier DEa in
Fig. 5a) to get adsorbed into the hydrophobic interlayer space.
In comparison, in the desorption process, it has to overcome
the hydrophobic interaction (energy barrier DEd in Fig. 5a) with
the lipid bilayer to enter the cells. As for short side-chain dyes,
they have a strong hydration effect but weak hydrophobic
interaction with the lipid bilayer, and so they are hard to adsorb
but easy to desorb into the cytoplasm. Inversely, long side-chain
dyes have a low hydration effect and strong hydrophobic
interaction, which results in easy adsorption but difficult
desorption. Thus, neither short nor long chains are favorable
for cellular delivery, while an appropriate length side-chain can
make both adsorption and desorption of dyes relatively easy,
thus enabling efficient cellular delivery.
Conflicts of interest
There are no conicts to declare.
Acknowledgements
Macroscopically, changing the side-chain length mainly
changed the log P of dyes. Thus, in order to further quantify the The study was supported by the National Natural Science
relationship between the side-chain length and cell perme- Foundation of China (51773111, 21672130, 51273107,
ability, the clog P values40 of SPs and 9E-BMVCs are presented in 51273175, and 51573158), Natural Science Foundation of
Tables S3 and S4.† For different compounds, the scope of clog P Shandong Province, China (ZR2017ZC0227), Fundamental
for efficient cellular delivery may be distinct, due to the Research Funds of Shandong University (2017JC011 and
substantial difference in the molecule structure. Concretely 2018JC030), and MOE Key Laboratory of Macromolecular
speaking, for SPs with a side-chain on one side, when the clog P Synthesis and Functionalization, Zhejiang University
is larger than 6.59, the dyes are stuck in the cell membrane and (2018MSF03).
This journal is © The Royal Society of Chemistry 2019
Chem. Sci.

View Article Online
Chemical Science
Edge Article
22 A. S. Klymchenko and R. Kreder, Chem. Biol., 2014, 21(1), 97–
113.
Notes and references
23 M. Tian, Y. Liu, Y. Sun, R. Zhang, R. Feng, G. Zhang, L. Guo,
X. Li, X. Yu, J. Z. Sun and X. He, Biomaterials, 2017, 120, 46–
56.
1 D. J. Stephens and V. J. Allan, Science, 2003, 300(5616), 82–86.
2 J. A. Conchello and J. W. Lichtman, Nat. Methods, 2005, 2(12),
920–931.
24 L. Guo, R. Zhang, Y. Sun, M. Tian, G. Zhang, R. Feng, X. Li,
X. Yu and X. He, Analyst, 2016, 141(11), 3228–3232.
25 S. R. Marder, J. W. Perry and W. P. Schaefer, Science, 1989,
245(4918), 626–628.
26 G. S. He, J. D. Bhawalkar, C. F. Zhao, C. K. Park and
P. N. Prasad, Appl. Phys. Lett., 1996, 68(25), 3549–3551.
27 G. S. He, Y. Cui, J. D. Bhawalkar, P. N. Prasad and
D. D. Bhawalkar, Opt. Commun., 1997, 133(1), 175–179.
28 G. S. He, L. Yuan, Y. Cui, M. Li and P. N. Prasad, J. Appl. Phys.,
1997, 81(6), 2529–2537.
3 I. Johnson, Histochem. J., 1998, 30(3), 123–140.
4 A. P. de Silva, H. Q. N. Gunaratne, T. Gunnlaugsson,
A. J. M. Huxley, C. P. McCoy, J. T. Rademacher and
T. E. Rice, Chem. Rev., 1997, 97(5), 1515–1566.
5 S. N. W. Toussaint, R. Calkins, S. Lee and B. W. Michel, J. Am.
Chem. Soc., 2018, 140(41), 13151–13155.
6 K. A. Mix, J. E. Lomax and R. T. Raines, J. Am. Chem. Soc.,
2017, 139(41), 14396–14398.
7 M. C. Morris, J. Depollier, J. Mery, F. Heitz and G. Divita, Nat.
Biotechnol., 2001, 19(12), 1173–1176.
29 Y. Chen, L. Qiao, L. Ji and H. Chao, Biomaterials, 2014, 35(1),
2–13.
30 S. W. Perry, J. P. Norman, J. Barbieri, E. B. Brown and
H. A. Gelbard, BioTechniques, 2011, 50(2), 98–115.
8 K. Johnsson, Nat. Chem. Biol., 2009, 5(2), 63–65.
9 P. L. Mcneil, Methods Cell Biol., 1988, 29, 153–173.
10 K. Barber, R. R. Mala, M. P. Lambert, R. Qiu,
R. C. Macdonald and W. L. Klein, Neurosci. Lett., 1996,
207(1), 17–20.
˜
31 M. Poot, Y. Z. Zhang, J. A. KraMer, K. S. Wells, L. J. Jones,
D. K. Hanzel, A. G. Lugade, V. L. Singer and
R. P. Haugland, J. Histochem. Cytochem., 1996, 44(12),
1363–1372.
11 R. Y. Tsien, Nature, 1981, 290(5806), 527–528.
12 J. M. Gillis and P. Gailly, Biophys. J., 1994, 67(1), 476–477.
13 K. G. Morgan, Biophys. J., 1993, 65(2), 561–562.
14 S. Du, S. S. Liew, L. Li and S. Q. Yao, J. Am. Chem. Soc., 2018,
140(47), 15986–15996.
32 R. J. Oliveira, P. C. Whitford, J. Chahine, V. B. P. Leite and
J. Wang, Methods, 2010, 52(1), 91–98.
¨
33 S. Wakayama, S. Kiyonaka, I. Arai, W. Kakegawa, S. Matsuda,
K. Ibata, Y. L. Nemoto, A. Kusumi, M. Yuzaki and
I. Hamachi, Nat. Commun., 2017, 8, 14850.
15 B. Ketterer, B. Neumcke and P. Lauger, J. Membr. Biol., 1971,
5(3), 225–245.
16 M. F. Ross, G. F. Kelso, F. H. Blaikie, A. M. James,
´
34 H. Y. Yuan, Y. M. Huang, J. D. Yang, Y. Guo, X. Q. Zeng,
S. Zhou, J. W. Cheng and Y. H. Zhang, Spectrochim. Acta,
Part A, 2018, 200, 330–338.
H. M. Cocheme, A. Filipovska, T. D. Ros, T. R. Hurd,
R. A. J. Smith and M. P. Murphy, Biochemistry, 2005, 70(2),
222–230.
35 M. Ge, P. Bai, M. Chen, J. Tian, J. Hu, X. Zhi, H. Yin and
J. Yin, Anal. Bioanal. Chem., 2018, 410(9), 2413–2421.
36 B. Glancy, L. M. Hartnell, D. Malide, Z. X. Yu, C. A. Combs,
P. S. Connelly, S. Subramaniam and R. S. Balaban, Nature,
2015, 523(7562), 617–620.
37 X. Li, M. Tian, G. Zhang, R. Zhang, R. Feng, L. Guo, X. Yu,
N. Zhao and X. He, Anal. Chem., 2017, 89(6), 3335–3344.
38 Molecular Probes™ Handbook, 11th edn, 2011.
39 N. Li, Y. Y. Liu, Y. Li, J. B. Zhuang, R. R. Cui, Q. Gong, N. Zhao
and B. Z. Tang, ACS Appl. Mater. Interfaces, 2018, 10(28),
24249–24257.
17 R. F. Flewelling and W. L. Hubbell, Biophys. J., 1986, 49(2),
531–540.
18 A. A. Raoof, M. Gudipatim, D. C. Bibby, S. W. Reingold,
S. Weinbach and A. Raoof, US pat., US2003036568-A1, 2003.
19 N. Stevens, N. O'Connor, H. Vishwasrao, D. Samaroo,
E. R. Kandel, D. L. Akins, C. M. Drain and N. J. Turro, J.
Am. Chem. Soc., 2008, 130(18), 7182–7183.
20 Y. Zhang, J. Wang, P. Jia, X. Yu, H. Liu, X. Liu, N. Zhao and
B. Huang, Org. Biomol. Chem., 2010, 8(20), 4582–4588.
21 H. Zhou, J. Tang, L. Lv, N. Sun, J. Zhang, B. Chen, J. Mao,
W. Zhang, J. Zhang and J. Zhou, Analyst, 2018, 143(10),
2390–2396.
40 M. Ishikawa and Y. Hashimoto, J. Med. Chem., 2011, 54,
1539–1554.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2019