Luminescent Silica Nanosensors for Lifetime Based Imaging ofIntracellular Oxygen with Millisecond Time ResolutionLongjiang Ding, Wei Zhang, Yinglu Zhang, Zhenzhen Lin, and Xu-dong Wang*

Department of Chemistry, Fudan University, 200433 Shanghai, P. R. China

*S Supporting Information

ABSTRACT: Intracellular oxygen concentration was quantitativelyimaged and rapidly traced with millisecond time resolution. Wehave demonstrated a new kind of oxygen nanosensors based on aruthenium complex doped solid silica nanoparticles, which showedhigh oxygen sensing performance (I0/I100 = 3.29, t95 < 3 s) and easeof surface functionalization. Their sensing performance can betuned by changing types of oxygen-sensitive probes and particlemorphology. The nanosensors showed excellent control in bothsensor size (from 30 to 200 nm), monodispersity, morphology,surface chemistry, and batch to batch consistency. Their uniformsize distribution and good biocompatibility made them suitable forintracellular studies. Because the sensor surface can be easilyfunctionalized with arbitrary units (such as transmembrane motifs,drugs, organelle-targeting groups, imaging reagent, and multiplesensor probes), these nanosensors provide a general platform to build easy-to-use tools for intracellular applications. The ease ofsurface functionalization was demonstrated by modifying the sensors outer surface with morpholinopropylamine and (3-carboxypropyl) triphenyl phosphonium, to actively target intracellular lysosomes and mitochondria of the tested cell lines(HeLa, MCF-7, and MCF-10A). Applying the mitochondria-targeting oxygen nanosensor together with our custom-built rapidphosphorescent lifetime imaging system, variations of intracellular oxygen have been quantitatively imaged and traced (inmillisecond intervals) in real time and in situ.

Intracellular metabolism of molecular oxygen is the majorroute to provide basic energy for cellular activities and

functionalities. Excess supply and insufficient delivery ofintracellular oxygen would result in both damage to cellbehavior and viability.1,2 Tracing variations of intracellularoxygen concentration and studying how cells (especially brain,neuronal, and immune cells) utilize oxygen provides invaluableinformation for biology and medicine. Thus, it is important tomonitor and quantify oxygen delivery, diffusion, anddistribution inside cells online and in situ.Oxygen sensing based on luminescence-quenching attracted

much attention in recent two decades due to their ease ofusage, noninvasiveness, capability of intracellular imaging (highthroughput), remote signal readout, and miniaturized sizedown to the nanosize range.3−5 It has become the majormethod of choice in studying intracellular oxygen variation.6

However, most oxygen-sensitive probes (OSPs) possess a largeconjugated π-system in their chemical structures, and arehydrophobic in nature, which make them difficult to bedirectly used for intracellular studies.5,7 Therefore, it is not asurprise to witness that encapsulating hydrophobic OSPs incompatible gas-permeable nanomaterials to form nanosensorbecomes the most popular approach for intracellular oxygensensing.8 Water-soluble OSPs have been reported andsuccessfully applied for intracellular studies.9−14 These probeshave much smaller size compared with polymeric nanosensors

and offer high spatial resolution. However, their applicationswere limited by complicated synthesis and difficulties in bothfunctionalization and cellular internalization. The hydrophobicluminescent core of these water-soluble probes should be fullyprotected, for example, by dendritic structure.15 Otherwise,their luminescence can be influenced by many factors,including solvent effect, ions, biomolecules, etc. Modificationof hydrophobic fluorinated OSPs with hydrophilic glucose andgalactose provided another solution.16 These conjugates hadminimum aggregation and self-quenching and were success-fully applied as oxygen imaging probes for 3D tissue models.However, their hydrophobic luminescent core was not fullyprotected and may still suffer from influences of potentialquenchers.Modern optical oxygen nanosensors were mostly prepared

using commercially available polymers, such as polystyr-ene,17−20 polyacrylamide,21,22 Eudragit RL-100,23,24 poly-(styrene-block-vinylpyrrolidone),25 luminescent conjugatedpolymer dots,26−30 and micelles,31−34 because they are readilyavailable with distinct oxygen permeability. However, polymerfunctionalization is difficult, since most polymers are chemi-

Received: August 15, 2019Accepted: October 25, 2019Published: October 25, 2019

Article

pubs.acs.org/acCite This: Anal. Chem. 2019, 91, 15625−15633

© 2019 American Chemical Society 15625 DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

Dow

nloa

ded

via

FUD

AN

UN

IV o

n D

ecem

ber

23, 2

019

at 0

4:54

:57

(UT

C).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

cally inert after production, and introduction of functionalgroups will influence both polymerization process andpolydispersity of obtained nanomaterials. Nanoprecipitationof binary polymer composites provides an alternative approachto produce nanosensors with chemically functional surface.35

However, it is nearly impossible to precisely control and tunesurface chemistry of polymer nanomaterials while keeping theirmonodispersity. Techniques for particle size control ofpolymer-based nanosensors are not well established, andmost of them always exhibit poor dispersity and broad sizedistribution. Established techniques for monodispersed poly-mer nanoparticles (such as poststaining, microemulsionpolymerization, dispersed polymerization, and mini-emulsionsolvent evaporation) are limited to several polymers (e.g.,polystyrene). Numerous in vivo researches have revealed thatthe size, surface chemistry and morphology of nanomaterialsdetermine their biological fate.36−39 Therefore, the biodis-tribution of polymer-based nanosensors is hardly to beanticipated and controlled. Although the organic−inorganichybrid nanomaterials developed by Kopelman and co-workersprovided improved surface functionalities, they still suffer fromtheir poor size control (100−600 nm in diameter),40 and theirhydrophobic shell makes further chemical modificationdifficult.41

Silica nanoparticles have excellent size and morphologycontrol (from several nanometer to micrometer) and well-established functionalization techniques for surface chemistrymodulation.42 They are optically transparent, highly stable andmonodispersed in aqueous solution, and have reproduciblecomposition. All these features made them attractive andsuitable to build nanosensors for intracellular applications.43,44

However, due to their dense structure and poor gasadsorption,45 they were excluded from constructing nano-sensor for oxygen. Attempts have been made to immobilizeoxygen-sensitive probes in silica gel (with large size 9.5−11μm) and bulky silica xerogel.46−48 Obviously, these materialsare too large in size for in vivo applications. Chu et al.49

reported that Pt(II) complex can be entrapped in core−shellsilica nanoparticles (average size 170 nm). Embedding theseparticles in sol−gel matrix resulted in a planar sensor film foroxygen. In the reported silica nanoparticle based oxygennanosensors,50−52 OSPs were either physically absorbed orchemically immobilized on silica surface rather than encapsu-lated into the interiors. Physically absorbed OSPs can leachout, and requires hydrophobic silica surface to firmly hold theOSPs, which hinders the biocompatibility of the nanosensors.Covalently immobilization of OSPs on silica surface requireschemical modification of OSPs, and immobilized OSPs cannotbe fully protected from potential quenchers.In an attempt to dope OSPs inside the interior of solid silica

nanoparticles, we have observed that OSPs-doped solid silicananoparticles show good response to molecular oxygen. Theynot only show good sensitivity and fast response to oxygen butalso exhibit excellent size and morphology control and goodbiocompatibility. Interestingly, oxygen-sensing performancecan be tuned by changing the types of OSPs and silicamorphologies. Because of their “clean” outer surface, thesenanosensors can be further functionalized with arbitrary unitsfor intracellular active-targeting and multiple purposes ofapplications. We have demonstrated the ease of functionaliza-tion by modifying the surface of nanosensors with organelle-targeting groups and used them for sensing oxygenconcentration at subcellular level. Owing to the long

luminescent lifetime of the nanosensors, we are able toimage intracellular oxygen concentration using our custom-built microscopic rapid lifetime imaging system. Thecombination of rapid lifetime imaging technique with solid-silica based nanosensors shows superior accuracy andreproducibility, which provides important tools for intracellularoxygen studies, and allows us quantitatively imaging andtracing intracellular oxygen concentration over time and in situwith millisecond time resolution.

■ EXPERIMENTAL SECTIONCharacterization. The morphology and size of nano-

sensors were characterized using a Field Emission Trans-mission Electron Microscopy (TEM, FEI Tecnai G2 F20 S-Twin). The concentration of oxygen was controlled by twomass-flow controllers (Red-Y, www.voegtlin.com, Switzerland).Luminescence spectra was recorded using a fluorospectrometer(Hitachi F-7000, Japan). Zeta potential was measured using aMalvern Zetasizer Nano ZS (Malvern Instruments Ltd.,Malvern, UK). Cell viability was studied on a Synergy H1Hybrid Multi-Mode Microplate Reader. Fluorescence imagesof HeLa cells were obtained using a wide-field fluorescencemicroscope (Leica DMi8, Germany). The filter cube sets areshown in Table 1.

The rapid phosphorescent lifetime imaging (rPLIM) systemconsists of a PI-MAX4 Gen III ICCD camera (PrincetonInstruments, U.S.A.), a fiber-optic laser (FC-ML, Changchun,China), and the Leica DMi8 wide-field fluorescence micro-scope. Signal synchronization of the laser and the camera werecontrolled by a built-in timing generator of the ICCD camera.The laser beam was guided using an optical fiber into themicroscope. The excitation pulse duration was set at 100 μs,and gate width was optimized at 8 μs. The first gate wasopened 600 ns after the laser was tuned off. Lifetime imageacquisition and data processing were performed using theLightField software (v6.3.1) from Princeton Instruments andMatlab (v2015b) from MathWorks.

Synthesis of Oxygen-Sensitive Nanosensors. Non-porous Ru(dpp)@silica nanoparticles (RuNPSiNPs) weresynthesized following a modified Stober method.53 Typically,815 μL of deionized water, 100 μL of ammonium hydroxide(28%; 0.14 M), and 8.4 mL of ethanol were mixed at roomtemperature. Then, different amounts of tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) bis(perchlorate) complex(Ru(dpp)3(ClO4)2, denoted as Ru(dpp)) were dissolved in300 μL of tetraethyl orthosilicate (TEOS; 0.14 M) and addedin the mixture. The solution was stirred in dark for 24 h. Theobtained brownish silica nanoparticles were harvested bycentrifuging at a g-force of 17 000g. The produced nanosensorswere washed with ethanol for at least three times to removeunbounded dyes and unreacted chemicals, and stored inethanol at a concentration of 10.0 mg mL−1.

Table 1. Filter Cube Sets for the Fluorescence Images Usingthe Wide-Field Fluorescence Microscopy

nameexcitation(nm)

emission(nm)

Ru(dpp) yellow channel 425−475 575−625lyso-tracker green green channel 340−380 435−485mito-tracker deep red red channel 595−645 665−715

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15626

Mesoporous Ru(dpp)@silica nanoparticles (RuMPSiNPs)were synthesized in an aqueous buffer solution with pH 7.00.54

Initially, 0.25 g of hexadecyltrimethylammonium bromide(CTAB) and 0.092 g of polyoxyethylene (20) cetyl ether (Brij-58) were dissolved in 25 mL of phosphate-buffered saline (pH7.00) at 95 °C under vigorous stirring. Then, 450 μL of TEOSwith dissolved Ru(dpp)3(ClO4)2 was added slowly in thebuffer solution. After continuous stirring for 8 h, synthesizednanoparticles were collected by centrifuging at 17 000g,washed three times by ethanol, and stored in ethanol.Dendritic mesoporous Ru(dpp)@silica nanoparticles (RuD-

PSiNPs) were prepared via the sol−gel approach.55 In a typicalprocess, 0.375 g of CTAB was mixed with 50 μL oftriethanolamine (TEOA) solution (TEOA/H2O (w/w) = 1/3), 7.5 mL of deionized water and stirred at 60 °C for 30 min.Subsequently, 2.0 mL of cyclohexane was added andcontinuously stirred for 5 min. Then 500 μL of TEOS withRu(dpp)3(ClO4)2 was added and stirred for 6 h. After that, themixture was added into 10 mL of ethanol, followed bycentrifuging at 17000g to isolate the particles. After washedwith ethanol for three times, the obtained RuDPSiNPs werestored in ethanol.Surface Modification with Organelle-Targeting

Groups. First, a solution of 3-morpholinopropylamine(MPA; 1.0 μM in ethanol) or (3-carboxypropyl)-triphenylphosphonium bromide (TPP-COOH; 1.0 μM inDMSO) was reacted with N,N′-dicyclohexylcarbodiimide(DCC; 3.75 μM) and N-Hydroxysuccinimide (NHS; 3.75μM) at room temperature for 2 h to give active-ester. Then, 10μmol 3-aminopropyltriethoxysilane (APTES) was added intothe active-ester solution. The mixture was kept in dry argonatmosphere, and stirred for 12 h to give silanes with active-targeting groups. Finally, the obtained silanes were added into1.0 mL 10.0 mg mL−1 RuNPSiNPs and incubated on a rotaryshaker for 20 h. The solid was collected by centrifuging at17000 g, followed by washing three times with ethanol andwater. The surface-modified RuNPSiNPs were dispersed inethanol at a concentration of 10.0 mg mL−1.Oxygen Responses of the Nanosensors Based on the

Measurement of Luminescent Lifetime. The oxygenresponse of the nanosensors was studied directly on themicroscope. Typically, 1.5 mL oxygen-sensitive nanosensorsdispersed in water at a concentration of 1.0 mg/mL was placedin a plastic microscopic dish. The solution was then saturatedwith synthetic gas containing different concentrations ofoxygen for 15 min. The oxygen concentration in the syntheticgas was adjusted using two mass-flow controllers. Thenanosensors were imaged on the Leica DMi8 fluorescentmicroscope equipped with the filter cube set for Ru(dpp)(Table 1). The lifetimes of nanosensors at differentconcentrations of dissolved oxygen were measured at roomtemperature using the rPLIM system. The detailed setup forthe rPLIM system was described in the characterizationsection. The measured luminescent lifetimes were plottedagainst the dissolved oxygen to obtain the calibration curve forintracellular measurements.Cell Culture and Cytotoxicity Assay. Throughout the

experiments, HeLa cells and MCF-7 cells were cultured inMEM media supplemented with 10% FBS, 1% L-glutaminesolution, 1% penicillin-streptomycin solution, and 1% MEMnonessential amino acid solution. MCF-10A cells wereincubated in the special MEM media (item number CM-0525) bought from Procell Life Science&Technology Co., Ltd.

(www.procell.com.cn). Cells were cultivated at 37 °C in ahumidified atmosphere with 5% CO2. Cytotoxicity of nano-sensors were determined using CCK-8 assay. Initially, HeLacells were seeded into a 96-well plate at a density of 3000 cellsper well and allowed to settle for 24 h. Then the culture mediawas discarded, and the fresh culture media (100 μL)containing different concentrations of nanoparticles (0, 50,100, 150, 200, and 250 μg mL−1) was added into each well.Five replicated experiments were performed for eachconcentration. After 24 h, culture medium suspended withfree nanosensors was discarded, and cells were washed threetimes with phosphate buffer solution. Then 100 μL culturemedium containing 10% (v/v) CCK-8 was added to each welland incubated for 2 h. The absorbance at 450 nm was recordedvia a multiwell plate reader, and the absorbance at 650 nm wasused as reference. Cell viability was calculated from thefollowing formula: cell viability (%) = the absorbance ofexperimental group/the absorbance of control group ×100%.

Intracellular Imaging. HeLa cells were incubated with200 μg mL−1 nanosensors for 4 h, then washed three timeswith the HBSS to remove free suspended nanosensors.Colocalization experiments were performed by incubatingcells with Lyso-Tracker Green (75 nM) or Mito-Tracker DeepRed (75 nM) for 1 h, respectively. The stained cells werewashed with PBS (pH 7.20−7.40) for three times to removefree dyes in the culture media. The Spearman’s rankcorrelation coefficient and the luminescence intensity of thefluorescence images was analyzed and calculated using ImageJsoftware. For tracing variations of intracellular oxygenconcentration, 30 μL glucose (1.0 M) and 20 μL glucoseoxidase (15 mg mL−1) solution were used to consumedissolved oxygen in the culture media.

■ RESULTS AND DISCUSSIONPreparation and Characterization of Nanosensors.

Silica nanoparticles are readily available in different particlesize, chemical composition and morphology with excellentproperty control and tunability.54,56,57 In order to encapsulateluminescent OSPs inside silica nanoparticles, OSPs shouldhave good solubility and chemically fit the physiochemicalproperties of silica matrix. There are residual silanol groups(−Si−OH) inside silica nanoparticle,58 which makes theinterior of silica negatively charged. To chemically fit thespecial microenvironment inside silica, we have selected apos i t i ve ly charged OSP, t r i s(4 ,7 -d ipheny l -1 ,10-phenanthroline)ruthenium(II) perchlorate (referred as Ru-(dpp)), to prepare nanosensors for oxygen.59,60 The Ru(dpp)dissolved well in silica precursor tetraethyl orthosilicate(TEOS), and directly encapsulated inside silica nanoparticlesduring particle formation. Attempts had been made toencapsulate photostable fluorinated metalloporphyrins (suchas PtTFPP) inside silica nanoparticles, but failed. Althoughthey were soluble in silica precursor TEOS, continuouslyhydrolization of TEOS produces hydrophilic oligomers andreduces the solution’s hydrophobicity. The reduction inhydrophobicity caused precipitation and aggregation ofPtTFPP. Therefore, condensation of hydrophilic oligomersand formation of nanoparticles occurred in the absence ofPtTFPP, which produced silica nanoparticles without oxygen-sensitive probes.The Ru(dpp)-doped nonporous solid silica nanoparticles

(RuNPSiNPs) were produced via the Stober method.Transmission electron microscopic (TEM) image (Figure

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15627

1a) showed the RuNPSiNPs were monodispersed and had anarrow size distribution (75 ± 5 nm, Figure S1). Thenanosensors showed excellent size control by simply adjustingammonia concentration in the synthetic media, and nano-sensors with diameter from 30 to 200 nm can be massivelyproduced (at gram level) via the one-pot reaction (Figure S2).Elemental mapping of RuNPSiNPs (Figure 1b) shows thatRu(dpp) were uniformly distributed throughout silica nano-particles. The RuNPSiNPs offer high luminescent brightnesseven in pure oxygen environment (Figure S3), suggestingsufficient number of photons emitted when excited at 445 nm.Calculation based on UV/vis absorption spectra reveals thatthere are ∼4680 Ru(dpp) molecules encapsulated inside asingle nanosensor, which warrants that it is bright enough formicroscopic imaging.The quenchability of RuNPSiNPs (expressed as I0/I100,

where I0 and I100 are the intensities at 0 and at 100% oxygen)reaches 3.29 (Figure 1c,d), which is even higher than that ofRu(dpp) in polystyrene particles (I0/I100 = ∼2.3),61 indicatingthe nonporous silica matrix have good oxygen permeability.The oxygen response is fully reversible, and the response time(t95) is less than 3 s (including gas diffusion time from the gascontrollers to the measurement cell). Moreover, theRuNPSiNPs showed excellent batch to batch consistency interms of size, morphology, monodispersity and oxygen sensingperformance, which ensures repeatable and reliable sensorproperties in real applications (Figure S4). A detailedcomparison of the new silica based oxygen nanosensor withreported intracellular oxygen sensors is summarized in Table

S1 in the Supporting Information, which shows the newoxygen nanosensor exhibits outstanding performance in almostall aspects.

Tunable Sensing Performance. The oxygen responsecan be tuned by modulating the morphology of thenanosensors. Two different kinds of Ru(dpp)-doped poroussilica nanoparticles have been synthesized. One is mesoporous(denoted as RuMPSiNPs), and the other is dendritic andporous (referred as RuDPSiNPs). Their oxygen responseswere studied and compared to that of nonporous nanosensors.Figure 2 shows these porous nanosensors have well-defined,uniform and regular pore structures, which are beneficial forgas molecules diffusing in and out. For RuMPSiNPs, TEMshowed their average diameter is 80 ± 5 nm (Figure 2a).Oxygen response study reveals RuMPSiNPs have a quench-ability of 2.14 (Figure 2b). The typical oxygen response time(t95) is only 2 s, which can be attributed to the porousstructure and high gas accessibility (Figure S5). Nanosensorsmade from dendritic porous silica nanoparticle had larger pores(∼10 nm). They were uniform in size and had an average sizeof 72 ± 8 nm (Figure 2c). The larger inner pores are beneficialfor the diffusion of oxygen molecules in and out, andRuDPSiNPs exhibited higher sensitivity, especially at lowoxygen concentration (< 20%; Figure 2d). Compared withnonporous nanosensors, the porous structure does notsignificantly improve sensor quenchability, which is out ofour expectation. Detailed studies have shown that the existenceof structure-directing surfactant influences oxygen diffusion(data not shown).

Figure 1. (a) TEM image of RuNPSiNPs (insert: the structure model). (b) EDX element mapping images of the elements Si, O, and Ru ofRuNPSiNPs. (c) The reversible oxygen response of the RuNPSiNPs. (d) The corresponding Stern−Volmer plot. The calibration curve fits wellwith the Stern−Volmer equation, and the two-site model was applied to calculate Stern−Volmer parameters. Stern−Volmer equation:II

fK

fK1 O1

1 O0 1 2 2 2= ++ * [ ]

−+ * [ ] .

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15628

The oxygen response can be further tuned by varying typesof OSPs. To chemically fit the microenvironment of silicananoparticles, hydrophilic OSPs are preferred and should bechemically immobilized inside silica nanoparticles to preventdye leakage. The hydrophilic probe, Pt(II) meso-tetra (4-carboxyphenyl) porphine (PtTCPP), was selected as anexample to show the capability of tuning sensor response.PtTCPP was covalently immobilized inside solid silicananoparticles using the same procedure as RuNPSiNPs.Results showed PtTCPP-doped silica nanosensors showmuch higher quenchability (I0/I100 = 18.67; Figure S6),because the Pt(II) porphyrins have much longer luminescence

lifetime. However, the poor photostability of PtTCPP limitedtheir applications in long-term intracellular imaging.

Measuring Dissolved Oxygen via LuminescenceLifetime. The long luminescence lifetime of OSP-doped silicananoparticles offers another attractive feature that intracellularoxygen can be imaged via lifetime measurement. Luminescencelifetime is an intrinsic parameter of a luminophore, and itsvalue is not depending on dye concentration and excitationlight intensity. Lifetime measurements are self-referenced, andimmune to photobleaching and autofluorescence, which arenotorious error sources for quantitative analysis, especiallyunder intense illumination on a fluorescence microscope. Theuse of highly stable luminescent materials62,63 together with

Figure 2. TEM images of RuMPSiNPs (a) and RuDPSiNPs (c), inserts are the structure models. The Stern−Volmer responses (two-site model) ofRuMPSiNPs (b) and RuDPSiNPs (d) at different concentrations of gaseous oxygen, respectively.

Figure 3. (a) Luminescence lifetime of RuNPSiNPs (10 mg mL−1) obtained via the rPLIM measurement at different concentration of dissolvedoxygen and room temperature and (b) the corresponding Stern−Volmer plot (two-site model).

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15629

ratiometric sensing can significantly improve measurementaccuracy when lifetime measurements are not available. Thelong luminescence lifetime of OSPs allows oxygen concen-tration can be measured using simpler equipment, becausemodulation of light source can be much slower, and brightLEDs can be used as light source. The excited-state lifetime ofRu(dpp) is a few microseconds.40,64 Encapsulation of multipleRu(dpp) molecules in one silica nanoparticle enhanced particlebrightness, which is beneficial for lifetime imaging. Usually,luminescence lifetime is measured via time correlated singlephoton counting technique (TCSPC). The fitting ofexponential decay curve gives the measured luminescencelifetime. However, this method is rather slow (requires severalminutes for a single-point measurement) for imaging, and isnot suitable for tracing fast intracellular events.In order to trace and accurate measure intracellular oxygen

concentration, we have, for the first time, adapted the rapidlifetime determination65 method for microscopic luminescencesensing and imaging, which we called rapid phosphorescentlifetime imaging (rPLIM). At optimized condition, the rPLIMapproach offers both high accuracy and excellent temporalresolution (up to tens of frames per second). The new rPLIMapproach can precisely measure luminescence lifetime ofRuNPSiNPs at different oxygen concentration (Figure S7),and the relative standard deviation with the TCSPC approachis only 0.39% (Figure S8). The acquisition rate is much faster,and it takes only 118 ms to image a sample at a spatialresolution of 1024 × 1024 pixels. The current system is limitedby the readout mode of ICCD camera, and can be even faster ifCMOS camera was used. In comparison, the TCSPC approachtakes almost 2 min to measure only one luminescence decay ata single spot, not to mention imaging an area using scanningmode. Alternative approaches for fast acquisition rate can beachieved using intensified CMOS camera66 or frequency-domain imaging system with high frame rate.67

As shown in Figure 3, the lifetime τ of RuNPSiNPsdecreases gradually with increasing concentration of dissolvedoxygen (Figure 3a). The calibration curve can be fitted wellusing the two-site model,68 which allows measurement ofdissolved oxygen in the range from 0 to 40 mg L−1 (Figure 3b).Since the rPLIM measurement is independent of probe

concentration, photobleaching, uneven distribution of dyesand light source intensity, detector sensitivity and autofluor-escence, it is an ideal tool for real-time and quantitativelytracing intracellular events.

Organelle-Targeted Imaging. Before applying theRuNPSiNPs for intracellular studies, their cytotoxicity inHeLa cells was studied using CCK-8 assay. Results showedthat cell viability remains above 90% when concentration ofnanosensors was below 150 μg mL−1 (Figure S9), demonstrat-ing their low cytotoxicity and good biocompatibility.Incubation of HeLa cells with RuNPSiNPs for 12 h showedthat bare RuNPSiNPs cannot pass through cell membranesand remain at extracellular space (Figure S10). This is mainlydue to the negatively charge outer surface (zeta potential,−41.2 mV) and results in low interaction with negativelycharged cell membranes.18,69

The well-established silane technique allows us to easilyfunctionalize nanosensors’ surface with required features. Wecan simply feature the nanosensors with lysosome- andmitochondria-targeting capabilities by covalently immobilizingMPA and TPP groups on nanosensor surface, respectively.Colocalization experiments showed that they had excellentorganelle-targeting capability, with Spearman’s rank correlationcoefficient above 0.8 for both lysosome and mitochondriatargeting (Figure 4 and S11). Figure 4a showed thatRuNPSiNPs-Lyso was mainly located in lysosome (Spearman’srank correlation coefficient, ρ = 0.81) and barely distributed inmitochondria (ρ = 0.06, Figure S11a). The RuNPSiNPs-Mitowas distributed mainly in mitochondria (ρ = 0.86, Figure 4b),and barely found in lysosomes (ρ = 0.3, Figure S11b).Moreover, unlike the polymer-based nanosensors whoseintracellular internalization is highly depending on celltypes,7 these new nanosensors showed uniform performancein different cell lines, including the tested breast cancer MCF-7cells (Figure S12a) and normal breast MCF-10A cells (FigureS12b). These results proved the ease of surface-functionaliza-tion of the new nanosensor, which provided an easy-to-useplatform for intracellular studies. The outer surface ofnanosensors can be further explored for intracellular multiplesensing, drug delivery, sense and treat of diseases, imagingguide surgeries, etc.

Figure 4. Intracellular localizations of RuNPSiNPs-Lyso (a) and RuNPSiNPs-Mito (b) in HeLa cells. Fluorescence images of RuNPSiNPs-Lyso(a1) and RuNPSiNPs-Mito (b1) collected in Yellow Channel. (a2) Fluorescence images of Lyso-Tracker Green collected in Green Channel. (b2)Fluorescence images of Mito-Tracker Deep Red collected in Red Channel. (a3 and b3) Merged images. (a4 and b4) Colocalization analysis images.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15630

The mitochondria-targeted nanosensors were further appliedfor tracing variations of intracellular oxygen concentration overtime. To better demonstrate the performance of the sensorsand the high speed and accuracy of the rPLIM approach, localoxygen concentration of HeLa cells was modulated usingglucose and glucose oxidase. When glucose oxidase catalyticallyoxidized glucose, oxygen was consumed and caused decrease inlocal oxygen concentration. Figure 5b (black line) showed thatthe brightness of nanosensors decreased significantly over time,because of dye photobleaching under intensive incident light(Figure S13). It is difficult to quantitatively measure oxygenconcentration via luminescence intensity (Figure 5a and redline in Figure 5b). In contrast, local intracellular oxygen

concentration can be measured using our rPLIM technique,which offers much more stable and reproducible data (Figure5c). As shown in Figure 5d, the luminescence lifetimeincreased rapidly and reached maximum after 5 min of addingglucose and glucose oxidase. The lifetime decreased again after5 min, because molecular oxygen in the air dissolved again inthe media and quenched sensor luminescence. After calculatingthe oxygen concentration according to the calibration plot(Figure S14), it is obvious that intracellular oxygenconcentration can completely recover to its initial value(around 8 mg mL−1, Figure 5e), which proves that the lifetimemeasurement is much more reliable and accurate. Because ofthe high brightness of the nanosensors, it takes only 118 ms to

Figure 5. (a) Luminescence intensity images of nanosensors inside HeLa cells at different time, dissolved oxygen was consumed using glucoseoxidase and glucose. (b) The luminescence intensity change of nanosensor over time inside HeLa cells. Lifetime images (c) and average values (d)of nanosensors inside HeLa cells at different time, the luminescence lifetime changed significantly over time. (e) The calculated dissolved oxygenconcentration traced at different time.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15631

record a microscopic lifetime image, which makes tracing rapidintracellular events (such as calcium burst, neuron activities)possible. The conventional laser scanning lifetime imagingtechniques need several minutes or even half of an hour toobtain a lifetime image, which will lose the temporalinformation and are not suitable for tracing fast intracellularevents. Owing to its good temporal resolution and highaccuracy, the combination of rPLIM with the new nanosensorsprovide important tools for intracellular oxygen studies.

■ CONCLUSIONIt is shown that OSPs-doped solid silica nanoparticles havegood sensitivity and fast fluorometric response to oxygen. Thesensors exhibit excellent control in size, morphology,monodispersity, surface chemistry, and batch to batchconsistency. Oxygen responses can be tuned by changingsilica morphologies and types of OSPs. Their large surface areaand ease of surface modification enable sensor surface to befunctionalized with arbitrary components (such as targeting,drug delivery, multiple sensing, imaging, etc.), which provide aversatile platform to build multifunctional nanosensors forintracellular applications. After modifying nanosensor surfacewith lysosome and mitochondria active-targeting groups, theycan be internalized by cells and precisely located in theseorganelles. By employing our custom-built rPLIM system,variations of intracellular oxygen concentration can be tracedin real time (with milliseconds interval) and quantitatively withsuperior accuracy and reproducibility. Because of its hightemporal resolution, this system is extremely suitable forstudying fast processes, such as photosynthesis, mitochondriafunction, and utilization of oxygen of neuron cells, response ofcellular metabolism to drugs, and so on.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge athttps://pubs.acs.org/doi/10.1021/acs.analchem.9b03726.

Source of materials, particle size distribution, lumines-cence spectra, size control, batch to batch consistency,luminescence lifetime, and cytotoxicity of RuNPSiNPs;response of the RuMPSiNPs; response of PtTCPPdoped solid silica nanoparticles; fluorescence images ofHeLa cells, MCF-7 cells, and MCF-10A cells treatedwith nanosensors; the photobleaching of the nano-sensors inside cells during the monitoring process andthe luminescence lifetime calibration plot (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: wangxudong@fudan.edu.cn.ORCIDXu-dong Wang: 0000-0002-3402-7995NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe acknowledge the National Key R&D Program of China(2017YFC0906800), National Natural Science Foundation ofChina (21775029), the Recruitment Program of GlobalExperts (1000 Talent program) in China, and the Program

for Professor of Special Appointment (Eastern Scholar) atShanghai Institutions of Higher Learning (No. TP2014004).

■ REFERENCES(1) Papkovsky, D. B.; Dmitriev, R. I. Chem. Soc. Rev. 2013, 42 (22),8700−8732.(2) Martinezcayuela, M. Biochimie 1995, 77 (3), 147−161.(3) Lee, J. H.; Lim, T. S.; Seo, Y.; Bishop, P. L.; Papautsky, I. Sens.Actuators, B 2007, 128 (1), 179−185.(4) Stetter, J. R.; Li, J. Chem. Rev. 2008, 108 (2), 352−366.(5) Wang, X. D.; Wolfbeis, O. S. Chem. Soc. Rev. 2014, 43 (10),3666−761.(6) Wolfbeis, O. S. BioEssays 2015, 37 (8), 921−928.(7) Dmitriev, R. I.; Papkovsky, D. B. Methods Appl. Fluoresc. 2015, 3(3), 11.(8) Schaferling, M. Angew. Chem., Int. Ed. 2012, 51 (15), 3532−3554.(9) Brinas, R. P.; Troxler, T.; Hochstrasser, R. M.; Vinogradov, S. A.J. Am. Chem. Soc. 2005, 127 (33), 11851−11862.(10) Zheng, X. C.; Wang, X.; Mao, H.; Wu, W.; Liu, B. R.; Jiang, X.Q. Nat. Commun. 2015, 6, 5834.(11) Sakadzic, S.; Roussakis, E.; Yaseen, M. A.; Mandeville, E. T.;Srinivasan, V. J.; Arai, K.; Ruvinskaya, S.; Devor, A.; Lo, E. H.;Vinogradov, S. A.; Boas, D. A. Nat. Methods 2010, 7 (9), 755−759.(12) Dirnagl, U. Nat. Methods 2010, 7 (9), 697−699.(13) Dmitriev, R. I.; Papkovsky, D. B. Cell. Mol. Life Sci. 2012, 69(12), 2025−2039.(14) Yoshihara, T.; Hirakawa, Y.; Hosaka, M.; Nangaku, M.; Tobita,S. J. Photochem. Photobiol., C 2017, 30, 71−95.(15) Lebedev, A. Y.; Cheprakov, A. V.; Sakadzic, S.; Boas, D. A.;Wilson, D. F.; Vinogradov, S. A. ACS Appl. Mater. Interfaces 2009, 1(6), 1292−1304.(16) Dmitriev, R. I.; Kondrashina, A. V.; Koren, K.; Klimant, I.;Zhdanov, A. V.; Pakan, J. M. P.; McDermott, K. W.; Papkovsky, D. B.Biomater. Sci. 2014, 2 (6), 853−866.(17) Wang, X. D.; Gorris, H. H.; Stolwijk, J. A.; Meier, R. J.; Groegel,D. B. M.; Wegener, J.; Wolfbeis, O. S. Chem. Sci. 2011, 2 (5), 901−906.(18) Wang, X. H.; Peng, H. S.; Yang, L.; You, F. T.; Teng, F.; Hou,L. L.; Wolfbeis, O. S. Angew. Chem., Int. Ed. 2014, 53 (46), 12471−12475.(19) Zhu, H. G.; McShane, M. J. J. Am. Chem. Soc. 2005, 127 (39),13448−13449.(20) Napp, J.; Behnke, T.; Fischer, L.; Wurth, C.; Wottawa, M.;Katschinski, D. M.; Alves, F.; Resch-Genger, U.; Schaferling, M. Anal.Chem. 2011, 83 (23), 9039−9046.(21) Clark, H. A.; Hoyer, M.; Philbert, M. A.; Kopelman, R. Anal.Chem. 1999, 71 (21), 4831−4836.(22) Lee, Y. E. K.; Ulbrich, E. E.; Kim, G.; Hah, H.; Strollo, C.; Fan,W. Z.; Gurjar, R.; Koo, S. M.; Kopelman, R. Anal. Chem. 2010, 82(20), 8446−8455.(23) Fercher, A.; Borisov, S. M.; Zhdanov, A. V.; Klimant, I.;Papkovsky, D. B. ACS Nano 2011, 5 (7), 5499−5508.(24) Kondrashina, A. V.; Dmitriev, R. I.; Borisov, S. M.; Klimant, I.;O’Brien, I.; Nolan, Y. M.; Zhdanov, A. V.; Papkovsky, D. B. Adv.Funct. Mater. 2012, 22 (23), 4931−4939.(25) Borisov, S. M.; Mayr, T.; Klimant, I. Anal. Chem. 2008, 80 (3),573−582.(26) Sun, K.; Tang, Y.; Li, Q.; Yin, S. Y.; Qin, W. P.; Yu, J. B.; Chiu,D. T.; Liu, Y. B.; Yuan, Z.; Zhang, X. J.; Wu, C. F. ACS Nano 2016, 10(7), 6769−6781.(27) Wu, C. F.; Bull, B.; Christensen, K.; McNeill, J. Angew. Chem.,Int. Ed. 2009, 48 (15), 2741−2745.(28) Dmitriev, R. I.; Borisov, S. M.; Dussmann, H.; Sun, S. W.;Muller, B. J.; Prehn, J.; Baklaushev, V. P.; Klimant, I.; Papkovsky, D.B. ACS Nano 2015, 9 (5), 5275−5288.(29) Zhao, Q.; Zhou, X. B.; Cao, T. Y.; Zhang, K. Y.; Yang, L. J.; Liu,S. J.; Liang, H.; Yang, H. R.; Li, F. Y.; Huang, W. Chem. Sci. 2015, 6(3), 1825−1831.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15632

(30) Lu, S. S.; Xu, W.; Zhang, J. L.; Chen, Y. Y.; Xie, L.; Yao, Q. H.;Jiang, Y. Q.; Wang, Y. R.; Chen, X. Biosens. Bioelectron. 2016, 86,176−184.(31) Wang, X. D.; Stolwijk, J. A.; Lang, T.; Sperber, M.; Meier, R. J.;Wegener, J.; Wolfbeis, O. S. J. Am. Chem. Soc. 2012, 134 (41),17011−17014.(32) Zheng, X. C.; Tang, H.; Xie, C.; Zhang, J. L.; Wu, W.; Jiang, X.Q. Angew. Chem., Int. Ed. 2015, 54 (28), 8094−8099.(33) McNamara, K. P.; Rosenzweig, Z. Anal. Chem. 1998, 70 (22),4853−4859.(34) Ji, J.; Rosenzweig, N.; Jones, I.; Rosenzweig, Z. Anal. Chem.2001, 73 (15), 3521−3527.(35) Mistlberger, G.; Koren, K.; Scheucher, E.; Aigner, D.; Borisov,S. M.; Zankel, A.; Polt, P.; Klimant, I. Adv. Funct. Mater. 2010, 20(11), 1842−1851.(36) Monopoli, M. P.; Aberg, C.; Salvati, A.; Dawson, K. A. Nat.Nanotechnol. 2012, 7 (12), 779−786.(37) Zhu, M. T.; Nie, G. J.; Meng, H.; Xia, T.; Nel, A.; Zhao, Y. L.Acc. Chem. Res. 2013, 46 (3), 622−631.(38) Huang, X. L.; Li, L. L.; Liu, T. L.; Hao, N. J.; Liu, H. Y.; Chen,D.; Tang, F. Q. ACS Nano 2011, 5 (7), 5390−5399.(39) Zhang, Z.; Wang, H.; Tan, T.; Li, J.; Wang, Z.; Li, Y. Adv. Funct.Mater. 2018, 28 (40), 1801840.(40) Xu, H.; Aylott, J. W.; Kopelman, R.; Miller, T. J.; Philbert, M.A. Anal. Chem. 2001, 73 (17), 4124−4133.(41) Koo, Y. E. L.; Cao, Y. F.; Kopelman, R.; Koo, S. M.; Brasuel,M.; Philbert, M. A. Anal. Chem. 2004, 76 (9), 2498−2505.(42) Wang, X. D.; Rabe, K. S.; Ahmed, I.; Niemeyer, C. M. Adv.Mater. 2015, 27 (48), 7945−50.(43) Korzeniowska, B.; Nooney, R.; Wencel, D.; McDonagh, C.Nanotechnology 2013, 24 (44), 442002.(44) Schulz, A.; McDonagh, C. Soft Matter 2012, 8 (9), 2579−2585.(45) Szekeres, M.; Toth, J.; Dekany, I. Langmuir 2002, 18 (7),2678−2685.(46) Acosta, M. A.; Ymele-Leki, P.; Kostov, Y. V.; Leach, J. B.Biomaterials 2009, 30 (17), 3068−3074.(47) Wolfbeis, O. S.; Leiner, M. J. P.; Posch, H. E. Microchim. Acta1986, 3 (5−6), 359−366.(48) Lei, B. F.; Li, B.; Zhang, H. R.; Lu, S. Z.; Zheng, Z. H.; Li, W.L.; Wang, Y. Adv. Funct. Mater. 2006, 16 (14), 1883−1891.(49) Chu, C. S.; Lo, Y. L.; Sung, T. W. Talanta 2010, 82 (3), 1044−1051.(50) Chan, C. M.; Chan, M. Y.; Zhang, M.; Lo, W.; Wong, K. Y.Analyst 1999, 124 (5), 691−694.(51) Choi, M. M. F.; Xiao, D. Anal. Chim. Acta 1999, 387 (2), 197−205.(52) Cheng, S. H.; Lee, C. H.; Yang, C. S.; Tseng, F. G.; Mou, C. Y.;Lo, L. W. J. Mater. Chem. 2009, 19 (9), 1252−1257.(53) Liang, J.; Lu, Z.; Xu, J.; Li, J.; Zhang, H.; Yang, W. J. Mater.Chem. 2011, 21 (4), 1147−1152.(54) He, Q.; Cui, X.; Cui, F.; Guo, L.; Shi, J. Microporous MesoporousMater. 2009, 117 (3), 609−616.(55) Liu, P. C.; Yu, Y. J.; Peng, B.; Ma, S. Y.; Ning, T. Y.; Shan, B.Q.; Yang, T. Q.; Xue, Q. S.; Zhang, K.; Wu, P. Green Chem. 2017, 19(23), 5575−5581.(56) Lee, Y. G.; Park, J. H.; Oh, C.; Oh, S. G.; Kim, Y. C. Langmuir2007, 23 (22), 10875−10878.(57) Stober, W.; Fink, A.; Bohn, E. J. Colloid Interface Sci. 1968, 26(1), 62−69.(58) Liu, W.; Liu, J.; Yang, X.; Wang, K.; Wang, Q.; Yang, M.; Li, L.;Song, C. J. Mater. Chem. B 2015, 3 (2), 323−329.(59) Bacon, J. R.; Demas, J. N. Anal. Chem. 1987, 59 (23), 2780−2785.(60) Ruggi, A.; van Leeuwen, F. W. B.; Velders, A. H. Coord. Chem.Rev. 2011, 255 (21), 2542−2554.(61) Hartmann, P.; Leiner, M. J. P.; Lippitsch, M. E. Anal. Chem.1995, 67 (1), 88−93.(62) Chu, B.; Wang, H.; Song, B.; Peng, F.; Su, Y.; He, Y. Anal.Chem. 2016, 88 (18), 9235−9242.

(63) Chu, B.; Song, B.; Ji, X.; Su, Y.; Wang, H.; He, Y. Anal. Chem.2017, 89 (22), 12152−12159.(64) Chu, C. S.; Lo, Y. L. Sens. Actuators, B 2010, 151 (1), 83−89.(65) Woods, R. J.; Scypinski, S.; Love, L. J. C. Anal. Chem. 1984, 56(8), 1395−1400.(66) Hirvonen, L. M.; Festy, F.; Suhling, K. Opt. Lett. 2014, 39 (19),5602−5605.(67) Koren, K.; Mosshammer, M.; Scholz, V. V.; Borisov, S. M.;Holst, G.; Kuhl, M. Anal. Chem. 2019, 91 (5), 3233−3238.(68) Fuller, Z. J.; Bare, W. D.; Kneas, K. A.; Xu, W. Y.; Demas, J. N.;DeGraff, B. A. Anal. Chem. 2003, 75 (11), 2670−2677.(69) Chung, T. H.; Wu, S. H.; Yao, M.; Lu, C. W.; Lin, Y. S.; Hung,Y.; Mou, C. Y.; Chen, Y. C.; Huang, D. M. Biomaterials 2007, 28 (19),2959−2966.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.9b03726Anal. Chem. 2019, 91, 15625−15633

15633