Dye-doped silica nanoparticle probes for ﬂ uorescence lifetime imaging of reductive environments in living cells †

a Fluorescence detection sensitivity can be drastically improved by the application of nanoparticles (NPs) because of their superior brightness compared to organic dyes. Here, using dye-doped silica NPs (SiNPs), we developed FRET-based nanoparticle probes for the detection of reductive environments in living cells. To this end, we designed three FRET acceptors based on black hole quenchers (BHQs). Their polarity was tuned by introducing hydroxyl, PEG and sulfate groups. To conjugate them to NPs, we used an original pre-functionalization approach, where the quencher was coupled by a “ click ” reaction to Pluronic F127 and further used for the preparation of silica NPs. This approach enabled easy preparation of silica NPs functionalized with varying amounts of quenchers by simple mixing of functionalized and parent Pluronic F127 in di ﬀ erent mol%. The increase in the quencher concentration at the SiNPs surface produced a rapid drop in the ﬂ uorescence intensity with 80% quenching and a 2-fold drop in the emission lifetime for 16 mol% of the quenchers. Then, to obtain turn-ON sensing of reductive environments, the quenchers were coupled to the NPs through a disul ﬁ de linker using the same pre-functionalization strategy. The obtained nano-probes showed a >10-fold increase in their ﬂ uorescence in the presence of reductive agents, such as tris(2-carboxyethyl)phosphine (TCEP) and glutathione. Remarkably, BHQ quencher bearing sulfate group showed the highest turn-ON response, probably due to its superior capacity to escape from the NP surface after disul ﬁ de bond cleavage. The obtained best nanoprobe was successfully applied for detection of reductive environments inside living cells using ﬂ uorescence lifetime imaging (FLIM). This work provides insights for FRET acceptor design and its controlled grafting, which enables preparation of the ﬁ rst redox-sensitive silica nanoparticle probe for lifetime imaging.


Introduction
2][3] They enable detection of bioactive molecules and monitoring of biomolecular processes with the highest speci-city and sensitivity.However, the ever present problem of uorescent molecular probes is their brightness, due to the limited values of their extinction coefficient and quantum yields. 46][7] So far, a variety of uorescent NPs has been developed: quantum dots (QDs), 8,9 dye-doped silica NPs (SiNPs), [10][11][12][13] nanodiamonds, 14 conjugated polymer NPs, 15,16 dyedoped polymer NPs, 7,[17][18][19] etc.Within this list, dye-loaded silica NPs present particular interest due to the high brightness and photostability they can reach, combined with the very low toxicity of many of their formulations.Moreover, their high versatility offered by the possible synthetic strategies and by the introduction of effective collective photophysical processes 18,20 makes them a powerful toolbox for addressing many different imaging and sensing applications in the eld of nanomedicine.
The most common mechanism in the design of uorescent probes is Förster Resonance Energy Transfer FRET. 213][24] In contrast, NPs are less efficient FRET donors or acceptors because frequently their size is much larger than the Förster radius.Thus, earlier works on quantum dots showed that efficient FRET requires $10 FRET acceptors at the periphery of the organic shell, 25,26 though there are examples for high FRET with fewer acceptors, but located just next to the inorganic core of NPs. 27Dye-doped silica NPs are particularly promising FRET donors for two reasons.At rst, recent protocols that use organic surfactant Pluronic as template enable preparation of ultrasmall NPs down to 10 nm size. 11Second, at high dye loading energy hopping within the encapsulated dyes can further improve the efficiency of the energy transfer. 20,28As we have demonstrated, this process, due to the energy transfer among chromophores of the same nature (the so-called "homo-FRET") plays in fact a prominent role in distributing the excitation energy among several donor dyes, including the closest ones to the nal acceptors. 20][40] Herein we describe the development of FRET-based nanosensors using rhodamine-doped silica nanoparticles (SiNPs) linked to amphiphilic BHQ-2 type quenchers through a reducible disulde (S-S) group.2][43] We varied the polarity of the quenchers by adding different polar heads and studied their interaction with uorescent SiNPs as well as their ability to quench their uorescence by direct adsorption and by covalent linkage.These nanosensors displayed a uorescence increase (turn-ON effect) upon reduction of the disulde bond by tris(2-carboxyethyl) phosphine (TCEP) or glutathione that separate the BHQ quencher from the particles.Moreover, this quenching system showed signicant effect on the uorescence lifetime of the SiNPs, which allowed us to use this system in Fluorescence Lifetime Imaging Microscopy (FLIM) experiments in live cells.

Design of the nanoprobes
We designed a system where a uorescent SiNP bears covalently linked quenchers through a disulde bond.Upon reduction of this bond, the quenchers separate from the particle and provoke a uorescence increase as the particle recovers its initial uorescence.As a particle we used previously reported rhodamine B-doped SiNPs, prepared using Pluronic F127 as a scaffold. 36These NPs feature $10 nm core diameter and $5 nm size PEG shell.In order to achieve an efficient turn-ON of these SiNPs, the quencher must full several properties.First, the overlap between the particle's emission spectrum and the absorbance spectrum of the quencher must be sufficient to insure an appropriate quenching effect.Therefore, Black Hole Quencher-2 (BHQ-2) was chosen as FRET acceptor.Then, the quencher should present an intermediate polarity, so that it could locate inside the PEG shell and thus close to the uorescent core of the SiNPs to ensure efficient FRET.On another hand, once the disulde bond get cleaved, the quencher moiety must be sufficiently hydrophilic so that it can escape from the particle in order to recover the uorescence of SiNPs providing an optimal turn-ON effect.However, BHQ quenchers are noncharged apolar molecules and therefore have a lipophilic nature that could be detrimental for their escape from SiNPs.We consequently designed and synthesised three different functionalizable and amphiphilic quenchers: BHQ-OH, BHQ-SO 3 and BHQ-PEG that bear, in addition to a "clickable" alkyne group, different polar heads, i.e., hydroxyl, sulphate and methoxy-PEG 8 groups, respectively (Fig. 1).
At the rst step of the synthesis of BHQ-OH, N-(2-hydroxyethyl)aniline was reacted with propargyl bromide to introduce a clickable group.The obtained aniline was reacted with Fast Black K giving rise to BHQ-OH (Scheme 1).BHQ-OH was efficiently converted to BHQ-SO 3 by sulfation using the SO 3 $NEt 3 complex with a yield of 87%.BHQ-OH was clicked to a methoxy PEG 8 moiety to afford compound 3.The latter was then converted into an active carbonate 4, which was coupled to propargylamine to obtain BHQ-PEG (Scheme 1).

Model FRET experiments
These amphiphilic quenchers were rst used to study their ability to interact non-specically with the surface of SiNPs.Increasing amounts of BHQs were added to an aqueous solution of rhodamine B doped SiNPs and the uorescence spectra were recorded.It was found that the quenching efficiency was dependent on the structure of the quencher (Fig. 2).At a ratio of two BHQ molecules per SiNP (10 rhodamine B dyes per SiNP, calculated as previously reported 36 ), BHQ-OH, BHQ-PEG and BHQ-SO 3 quenched 30, 40 and 55% of the SiNP uorescence, respectively.At 10 quenchers per NP, nearly 80% of the SiNP uorescence was quenched for all three quenchers.Thus, all three quenchers displayed capacity to bind SiNPs non-specically, though this behaviour depended slightly on the nature of the polar group.

FRET to covalently graed quenchers
In a second step, the quenching ability of these amphiphilic BHQs, being covalently attached to the surface of SiNPs, was investigated.In one approach, rhodamine B doped SiNPs bearing an azide group at the terminal end of the Pluronic F127 (Pluronic) were prepared. 44The functionalization of the SiNPs with the clickable quenchers by CuAAC "click" chemistry was not efficient, as we could not see signicant differences between the samples with and without Cu-catalyst.This could be explained by the poor efficiency of the "click" reaction at low concentrations in the presence of oxygen.Higher efficiency would require the use of a special chelated Cu(I) catalyst. 45In a second approach, Pluronic-N 3 was directly functionalised with the quenchers by CuAAC click chemistry prior to the formulation of SiNPs.The reaction proceeded smoothly and provided pure bi-functionalised Pluronic F127 (Scheme 2) according to NMR.Then, the modied Pluronic polymer was used to prepare SiNPs using the same protocol.Previously, it was estimated that a single SiNP was constituted of $50 Pluronic polymers, 46,47 which corresponded to $100 functionalizable sites per particle.Therefore, SiNPs bearing 0, 2, 4, 8 and 16 BHQs at their surface were readily obtained by mixing Pluronic and Pluronic-BHQ in appropriate ratios prior to formulation, i.e. 0, 2, 4, 8 and 16% of Pluronic-BHQ in total Pluronic, respectively.Dynamic light scattering (DLS) measurements showed that the Pluronic-BHQ molecules did not affect the size (24 nm) and the low polydispersity value (typically <0.1) of the obtained SiNPs (Table 1).This important result shows that Pluronic modied with quencher can be readily used as a building block to prepare functionalized SiNPs with preserved size characteristics.
The increase in the fraction of Pluronic-BHQ during particle synthesis resulted in the gradual increase in the absorbance of the obtained SiNPs as well as broadening of the absorption band with appearance of a shoulder above 600 nm (Fig. 3A).These spectral changes correspond to the increased contribution of the absorption of the BHQ quenchers, which conrms their successful graing to SiNPs.Fluorescence measurements of the obtained SiNPs-BHQs showed that an increase in the number of BHQ molecules per particle leads to a gradual decrease of the uorescence intensity (Fig. 3B and C).Despite their different polar groups, the quenching efficiency was similar for all three quenchers, so that 16 BHQ molecules per particle quenched more than 80% of SiNPs uorescence in all cases (Fig. 3C).The average uorescence lifetime also showed a gradual decrease upon an increase in the number of BHQ quenchers (Fig. 3D), though this effect was slightly stronger for BHQ-OH.Overall, the lifetime decreased from $3.7 ns in the parent SiNPs up to 1.9-2.1 ns in SiNPs bearing 16 quenchers per particle.It should be noted that the decrease in the lifetime is smaller than the decrease in the uorescence intensity, namely, <2-fold and $5-fold, respectively.This discrepancy indicates that FRET is probably not the only mechanism of quenching of  SiNPs by BHQ.We expect that other mechanisms (electron transfer and excitonic interactions) could produce dark species that decrease the uorescence intensity without detectable change in the lifetime measurements.Moreover, we cannot exclude a non-homogeneous distribution of dyes and/or quenchers in SiNPs, which could also affect differently uorescence intensity and the lifetime.Clearly, these experiments show that NPs could be efficiently quenched by a relatively small number of BHQ quenchers and this process can be followed by both intensity and lifetime measurements.

Preparation of nanoprobes and their validation in solution
Encouraged by the developed synthetic methodology and the quenching efficiency of the new amphiphilic BHQs, we prepared a uorogenic nanoprobe.The new BHQs quenchers were attached to the terminal PEG chains of the Pluronic F127 through a linker containing a disulde bond.For this purpose, bromopropanol was converted into alcohol azide 5, which was activated by carbonyldiimidazole 6.The latter was coupled to cystamine to give an amino-azide linker 7. Pluronic F127 was converted to a diactive carbamate 8 that was subsequently reacted with 7 to afford the Pluronic azide derivative 9.The latter was clicked to the 3 different amphiphilic BHQs (Scheme 3).The reducible Pluronicquencher conjugates were then mixed with Pluronic F127 in ratios 4/21 and 8/17 prior to the formulation to obtain SiNPs with respectively 16 and 32 BHQs at their surface.These ratios were chosen to insure an efficient quenching and therefore an enhanced turn-ON effect upon reduction.
As expected, the uorescence intensity decreased many-fold for all three quenchers and the effect was particularly strong with 32% of quenchers per particle (Fig. S1 †).To evaluate the ability of these nano-constructs to sense a reductive environment, they were incubated with 20 mM TCEP (pH was adjusted to 7.4 by sodium hydroxide) or 40 mM glutathione and their uorescence intensity was monitored over one hour.For all quenchers used at 32%, TCEP produced a strong increase in the uorescence intensity on the time scale of 1 h (Fig. 4A).The uorescence increase was lower for SiNPs with lower % of the quencher (16%) and no increase was observed for SiNPs without graed quencher (Fig. 4B, S2 and S3 †).Addition of glutathione also produced uorescence increase when the quencher was graed to SiNPs, though the uorescence intensity increased slower, probably because glutathione is a weaker reducing agent than TCEP (Fig. S4 †).Thus, for SiNPs with 32% BHQ-SO 3 in presence of 40 mM glutathione, the uorescence intensity reached a plateau only aer 10 h with a 12-fold enhancement (Fig. S5 †), which is close to the value obtained in the presence of TCEP in 1 hour (Fig. 4A).The uorescence intensity of the nanosensors (with S-S bond) in the absence of reducing agent was nearly invariant for at least 20 min, which demonstrated their stability in the phosphate buffer medium (Fig. S6 †).An invariant uorescence over time was also observed for BHQ-free SiNPs in the presence of glutathione, thus, excluding a direct effect of the reducing agent on SiNPs (Fig. S7 †).Finally, in our control SiNPs-BHQ-OH without reducible linker, no uorescence increase was observed in the presence of glutathione (Fig. S8 †).Taken together, these results suggest that the reducing agents cleave the S-S bonds, which leads to the liberation of the quenchers and thus to the increase in the SiNPs uorescence.
The three nanosensors (at 32% of graing) displayed signicantly different turn-ON effect efficiencies upon reduction as their uorescence enhancement varied from 3-to 13-fold aer one hour in the presence of TCEP (Fig. 4A) and 1.5 to 3.5fold aer one hour with glutathione (Fig. S4 †).The following ranking of BHQ efficiency in terms of turn-ON effect for the nanosensors could be drawn up: BHQ-SO 3 > BHQ-OH > BHQ-PEG.This tendency can be explained by the fact that, in these systems, an efficient turn-ON effect is ensured by three synergetic factors: (1) the accessibility of the S-S bond; (2) the capacity of BHQ to quench NPs and (3) the capacity of the BHQ to escape from SiNPs aer S-S bond reduction.For all three quenchers we expect that the access to S-S bond to the reductive agent should be similar.Moreover, all three quenchers at high percentage of graing showed similar quenching efficiency.Therefore, the observed differences are probably related to the differences in desorption of the quenchers from the SiNPs aer S-S bond reduction.The presence of a negative charge brought by the sulfate group of BHQ-SO 3 is likely favorable to the dissociation of the BHQ from the uorescent SiNP, that have a negative zeta potential, leading to an efficient de-quenching of this latter.By contrast, BHQ-PEG and BHQ-OH are noncharged which can explain their lower efficiency compared to BHQ-SO 3 due to less efficient desorption from SiNPs aer S-S bond reduction.More unexpected was the signicant difference in efficiencies between BHQ-PEG and BHQ-OH.Indeed BHQ-OH was shown to be >2 times more efficient than its PEGylated version despite the fact that the latter bears two triazole moieties as well as a PEG 8 which enhance its polarity.

Validation in living cells
The efficient turn-ON effect observed in the SiNP nanoprobes bearing 32% BHQ-SO 3 led us to investigate their ability to sense the reductive environment in living cells.HeLa cells were incubated for 2 hours in the presence of SiNPs with no quencher and SiNPs with 32% of BHQ-SO 3 .In both cases the cells showed intracellular dotted uorescence, which can be ambiguously assigned to internalized SiNPs (Fig. 5).The internalization of these nanoprobes probably took place by endocytosis, as suggested by the earlier data using analogous SiNPs. 48s SiNPs with 32% BHQ-SO 3 are very poorly uorescent in the intact form, the observation of uorescence inside the cells comparable to that of control NPs without graed quencher indicates that the internalized nanoprobes probably turned ON their uorescence due to the reductive environment present in cytosol and endosomes of living cells. 41,49However, as intensity response cannot provide denite answer regarding the particle turn-ON, we performed Fluorescence Lifetime Imaging Microscopy (FLIM), which can provide direct information about changes in FRET. 50,51n our FLIM studies, we rst compared the images of suspensions of control SiNPs and SiNPs bearing 32% of BHQ-SO 3 quenchers (Fig. 6).They showed clearly different pseudocolors (blue and yellow-green), with average lifetimes around 2.9 and 1.8 ns.Thus, in line with our time-resolved studies (see above), the BHQ-bearing SiNPs showed signicantly shorter lifetime due to FRET quenching process.Then, we incubated these two types of SiNPs with HeLa cells for 2 h.The obtained FLIM images showed strong intracellular uorescence in form of dots, indicating good internalization of both types of SiNPs (Fig. 6).However, in the case of SiNPs bearing BHQ-SO 3 , the pseudo-color showed clear heterogeneity (Fig. 6D), ranging between long lifetime of control SiNPs (blue) and the short lifetime of SiNPs with 32% of BHQ-SO 3 (yellow-green).This result suggests that aer internalization, some part of the nanoprobes underwent reductive cleavage at the S-S bond, which produced de-quenched SiNPs with lifetime close to control SiNPs.Thus, FLIM images validated the response of our nanoprobes to the reductive stimulus in the living cells.Moreover, we showed that FLIM imaging applied to FRET  nanoprobes is complementary to intensity measurements, because FLIM provides information about FRET efficiency, thus describing the probe response to the reductive environment.

Materials and methods
Milli-Q water (Millipore) was used in all experiments.All starting materials for synthesis were purchased from Alfa Aesar, Sigma Aldrich or TCI Europe and used as received unless stated otherwise.Synthesis of the BHQ and the Pluronic F127 derivatives are described in the ESI.† NMR spectra were recorded on a Bruker Avance III 400 MHz spectrometer.Mass spectra were obtained using an Agilent Q-TOF 6520 mass spectrometer.DLS measurements were performed on a Zetasizer Nano series DTS 1060 (Malvern Instruments S.A.).Absorption and emission spectra were recorded on a Cary 400 Scan ultraviolet-visible spectrophotometer (Varian) and a FluoroMax-3 spectrouorometer (Horiba Jobin Yvon) equipped with a thermostated cell compartment, respectively.For standard recording of uorescence spectra, the excitation wavelength was set to 530 nm and emission was recorded from 540 to 700 nm.The solutions of TCEP and glutathione were prepared in 20 mM phosphate buffer solutions (pH 7.4) and the pH was readjusted to 7.4.

Cell culture
HeLa cells (ATCC CCL-2) were grown in DMEM (Gibco-Invitrogen), supplemented with 10% fetal bovine serum (Lonza) and 1% antibiotic solution (penicillin-streptomycin, Gibco-Invitrogen) at 37 C in humidied atmosphere containing 5% CO 2 .Cells were seeded onto a chambered coverglass (IBiDi) at a density of 5 Â 10 4 cells per well 24 h before the microscopy measurements.For imaging, the culture medium was removed and the attached cells were washed with Opti-MEM (Gibco-Invitrogen).Next, a freshly prepared solution of NPs (50 nM) in Opti-MEM was added to the cells, which were incubated for 2 or 24 h.The medium was then removed and washed 2 times with HBSS and replaced by HBSS for imaging.Cell plasma membrane staining with wheat-germ agglutinin-Alexa488 (50 nM) was done at room temperature 10 min before the measurements.

Fluorescence lifetime measurements in solution
Time-resolved uorescence measurements were performed with the time-correlated, single-photon counting technique using the excitation pulses at 480 nm provided by a pulsepicked frequency doubled Ti-sapphire laser (Tsunami, Spectra Physics) pumped by a Millenia X laser (Spectra Physics).The emission was collected through a polarizer set at the magic angle and an 8 nm band-pass monochromator (Jobin-Yvon H10) at 582 nm.The single-photon events were detected with a microchannel plate photomultiplier (Hamamatsu) coupled to a pulse pre-amplier HFAC (Becker-Hickl GmbH) and recorded on a time correlated single photon counting board SPC-130 (Becker-Hickl GmbH).The instrumental response function (IRF) was recorded using a polished aluminum reector, and its full-width at half-maximum was $40 ps.Experimentally measured uorescence decays were deconvoluted with the instrumental response function and tted to retrieve the most probable lifetime distribution using the maximum entropy method (Pulse 5 soware). 52,53In all cases, the c 2 values were close to 1, indicating an optimal t.

Fluorescence lifetime imaging microscopy
Fluorescence microscopy experiments were performed on a home-built two-photon laser scanning setup based on an Olympus IX70 inverted microscope with an Olympus 60Â 1.2 NA water immersion objective. 54Two-photon excitation was provided by a Ti-sapphire (Tsunami, Spectra Physics) or an Insight DeepSee (Spectra Physics) laser.Imaging was carried out using two fast galvo-mirrors operating in the descanned uorescence collection mode.Photons were detected with an avalanche photodiode (APD SPCM-AQR-14-FC, PerkinElmer) which was connected to a time-correlated single photon counting (TCSPC) module (SPC830, Becker and Hickl) operating in reversed start-stop mode.The typical acquisition time was 60 s with an excitation power around 5 mW (830 nm) at the laser output.Subsequent data analysis using the soware SPC Image (Becker and Hickl) allowed us to extract the uorescence lifetimes from the decays, and their visualization in a FLIM image, using an arbitrary color scale.

Conclusions
Preparation of FRET-based sensors using nanoparticles remains a challenge because particle size is generally larger than the Förster distance.The problem can be solved by using very small NPs and a large number of quencher moieties.Here, we designed a nanosensor where uorescent dye-doped silica NPs (SiNP) bear covalently linked quenchers through a disulde bond.We prepared $20 nm silica NPs doped with rhodamine B dyes and graed with BHQ-2 type quenchers.It was found that optimal quenching (>10-fold) requires 16 to 32 quenchers at the SiNP surface.Moreover, the amphiphilic nature of the quencher was found to play a key role in the design of these nanosensors.Timeresolved uorescence measurements suggested in addition to the expected FRET mechanism, the presence of dark species due to static quenching.In the presence of TCEP or glutathione, SiNPs strongly increase their uorescence due to the reduction of the disulde bonds and separation of the quenchers from the particle.Laser scanning confocal imaging showed that particles can be internalized in cells.Fluorescence lifetime measurements (FLIM) revealed populations of internalized NPs with short and long lifetimes corresponding to weakly and strongly reduced SiNPs, respectively.Our data conrmed that the SiNPs nanoprobes lose their quenchers in response to the intracellular reductive environment.This work provides insights into design of FRET-based nanoparticle probes and demonstrates suitability of FLIM for characterization of these nanoprobes in living cells.

Fig. 1
Fig. 1 Structure of the 3 amphiphilic functionalizable BHQs.Polar heads are highlighted in red, the functionalizable site is highlighted in blue.

Fig. 3
Fig. 3 Spectroscopic properties of SiNPs with grafted BHQ quenchers.Absorption spectra (A) and emission spectra (B) of SiNP-BHQ-SO 3 (200 nM) with various amounts of BHQ per particle.Excitation wavelength was 530 nm.(C and D) Quenching effect of the covalently grafted BHQ quenchers on the fluorescence intensity (C) and the average lifetime (D) of SiNPs.

Fig. 4
Fig. 4 Effect of TCEP on the fluorescence of the fluorogenic SiNPs.(A) Fluorescence increase as a function of time of BHQ-SO 3 SiNPs (20 nM) in the presence of 20 mM TCEP in phosphate buffer 20 mM pH 7.4.(B) Fluorescence increase as a function of time of SiNPs bearing 32% BHQ-OH (red), BHQ-SO 3 (black) and BHQ-PEG (green) (20 nM) in the presence of 20 mM TCEP in phosphate buffer 20 mM pH 7.4.Excitation and emission wavelengths were 530 nm and 580 nm, respectively.

Fig. 6
Fig. 6 Intensity (A and B) and FLIM (C and D) images of SiNP-0% BHQ (A and C) and SiNP-32%-SS-BHQ-SO 3 (B and D) incubated at 50 nM for 2 h in the presence of HeLa cells.Insets of (C) and (D) at the top left corner show the FLIM image of corresponding SiNPs without (SiNP-0% BHQ) and with BHQ quencher (SiNP-32%-SS-BHQ-SO 3 ) in solution.Scale bar is 10 mm.

Table 1
Hydrodynamic diameter and polydispersity of the different SiNPs as measured by DLS