Mechanodelivery of nanoparticles to the cytoplasm of living cells†

Nyssa T. Emerson,a Chih-Hao Hsia,a‡ Ilona U. Rafalska-Metcalf,a and Haw Yang∗a

Received Xth XXXXXXXXXX 20XX, Accepted Xth XXXXXXXXX 20XXFirst published on the web Xth XXXXXXXXXX 200XDOI: 10.1039/b000000x

Nanotechnology has opened up the opportunity to probe,sense, and manipulate the chemical environment of biolog-ical systems with an unprecedented level of spatiotemporalcontrol. A major obstacle to the full realization of thesenovel technologies is the lack of a general, robust, and sim-ple method for the delivery of arbitrary nanostructures tothe cytoplasm of intact live cells. Here, we identify a newdelivery modality, based on mechanical disruption of theplasma membrane, which efficiently mediates the deliveryof nanoparticles to the cytoplasm of mammalian cells. Weuse two distinct execution modes, two adherent cell lines,and three sizes of semiconducting nanocrystals, or quan-tum dots, to demonstrate its applicability and effective-ness. As the underlying mechanism is purely physical, weanticipate that such “mechanodelivery” can be generalizedto other modes of execution as well as to the cytoplasmicintroduction of a structurally diverse array of functionalnanomaterials.

Owing to their unique photoluminescence (PL) or scatteringproperties, many nanomaterials have found use as comple-mentary optical probes to fluorophores and fluorescent pro-teins in the investigation of biological systems. Semiconduct-ing nanocrystals, or quantum dots (QDs), nanodiamonds, andmetallic nanomaterials1 have been used as reporters in im-munofluorescence,2,3 cell or protein tracking,4,5 and pH,6,7

temperature,8–10 or chemical11 sensing applications. Further-more, a myriad of functional nanomaterials have been devel-oped to not only probe but manipulate the cellular microenvi-ronment with unprecedented spatiotemporal control. For ex-ample, silica or protein-based delivery vehicles can be trig-gered to release chemical stimuli on demand,12–15 and mag-netic nanoparticles can be conjugated to biomolecules andmoved to locally initiate gene expression16 or tubulin poly-merization.17 These tools will no doubt become integral for

† Electronic Supplementary Information (ESI) available: Characterizationof QD diameter, passivation of QDs, electroporation protocol, flow cytomet-ric data analysis, and additional epifluorescence images of QD labeled cells,including Table S1 and Figs. S1–S15. See DOI: 10.1039/b000000x/a Department of Chemistry, Princeton University, Princeton, NJ 08544, USA.E-mail: hawyang@princeton.edu‡ Present address: Research Center for Applied Sciences, Academia Sinica,Nankang, Taipei, Taiwan.

elucidating the complex networks of events which unfold inthe intracellular space.

Yet much of the intracellular space is inaccessible to re-searchers due to the persistent difficulty of delivering arbitrarynanomaterials to the cytoplasm. Typically, without electri-cal, chemical, or mechanical intervention, exogenous mate-rials passively enter cells through endocytosis and remain se-questered in vesicles. Although many techniques have beendeveloped to deliver small molecules, proteins, and nucleicacids to the cytoplasm,18 most do not translate directly tonanomaterial delivery. For example, electroporation and lipo-fection are commonly used to deliver nucleic acids, but theyonly can deliver aggregates of QDs.19 As a result, both tech-niques are impractical for experiments that require nanomate-rials to be conjugated to specific intracellular proteins or struc-tures. Thus new delivery modalities are required to specifi-cally address the problem of nanomaterial delivery.

Many nanomaterial delivery modalities have been pro-posed,20,21 including chemical loading vectors, such as pep-tides22 or polymers;23,24 mechanical stress techniques, suchas microinjection;19 and fabricated devices, such as nanowirearrays25 or microfluidic channels for non-adherent cells.26

All of these techniques have successfully delivered nanoma-terials to the cytoplasm; however, their widespread adop-tion is hindered by a lack of either generality or accessibil-ity. For example, chemical loading vectors must be com-plexed with the nanoparticle surface to be effective, and itis unclear whether this can be performed on diverse materi-als. In contrast, microinjection can deliver almost any probeof interest, regardless of material composition, to the cyto-plasm; however, the procedure is labor intensive, extremelylow-throughput and often lethal. Many of these obstacles areovercome by fabricated devices, which are equally general andsubstantially improve on throughput and cell viability whencompared with microinjection. Unfortunately, because theirfabrication and/or operation requires sophisticated equipmentand expertise, at this point they remain largely inaccessibleto general research groups. A nanomaterial delivery modalitywhich is both general and accessible has remained elusive; thecommunity would benefit greatly from having such a deliverymodality.

A review of the currently available methods for cytoplasmic

1–7 | 1

plasmamembrane

nanoparticle

cytoplasm

nucleus

FF

F

recovery

mechanicalforces

c

ba

cells

30-gaugeneedle

glass bead

Fig. 1 Mechanodelivery concept. (a) Nanoparticles are typ-ically excluded from the cytoplasm. During mechanodelivery,generic mechanical force is employed to disrupt the plasmamembrane, allowing nanoparticles to diffuse through mem-brane wounds. Membrane wounds heal rapidly through en-dogenous mechanisms. Two specific implementations, (b)bead- and (c) scratch-loading are described in the text.

nanoparticle delivery leads us to reason that a delivery methodbased on mechanical forces will be general for nanoparticlesand nanostructures because it will be independent of both sur-face chemistry and biological responses (see Fig. 1a). Theremaining criterion is accessibility, ideally the simpler the ex-ecution method the better because then more practioners inthe community can benefit from such a method. A literaturesurvey brings us to a suite of elegant techniques, originally in-troduced by McNeil and co-workers over two decades ago.27

These methods use glass beads, narrow-gauge needles, orother implements to tear small, survivable holes in the plasmamembrane. Notably, these techniques could routinely deliverfluorescently-labeled dextran to the cytoplasm of mammaliancells. The simplicity makes these methods extremely appeal-ing; however, it is not apparent whether they can be directlytranslated to nanomaterial delivery, or if, as with electropora-tion, they are primarily applicable to soft macromolecules.

Here, we generalized two distinct mechanical stress tech-niques, bead-28 and scratch-loading,29 to delivering nanoma-terials to the cytoplasm of adherent mammalian cells. To per-form bead-loading, cells were grown on a glass coverslip andbombarded with ∼250-µm glass beads (see Fig. 1b). Alter-nately, scratch-loading was performed by repeatedly dragginga 30-gauge needle across the cell monolayer (see Fig. 1c).Although they differ radically in execution, both mechani-cal actions create transient disruptions in the plasma mem-brane which enable exogenous materials to diffuse into thecytoplasm. These mechanical disruptions heal very rapidlyin a physiological, calcium-mediated process; for example,micron-sized disruptions in the plasma membrane of Swiss-3T3 fibroblasts heal within 10 to 120 s,30 depending on the

scratch

site

c

NIH/3T3 WPE1-NB11

+ scratch

da b

NIH/3T3 WPE1-NB11

+ bead

V

C

e

Fig. 2 Cytoplasmic delivery of QDs by bead- and scratch-loading. (a-d) False color epifluorescence (top) and brightfield(bottom) images of live cells were acquired three hours aftertreatment. Trypan blue was included in the imaging mediumto quench PL arising from extracellular QDs. NIH/3T3 murinefibroblasts (a) and WPE1-NB11 prostate epithelial cells (b)bead-loaded with 1 µM QDs. NIH/3T3 (c) and WPE1-NB11(d) cells scratch-loaded under identical conditions. SuccessfulQD delivery is evidenced by the observation of diffuse cyto-plasmic PL (e.g. arrow C in b), in contrast with vesicular PL,which most likely results from endocytic uptake (e.g. arrow Vin b). (e) Close up epifluorescence (left) and brightfield (right)images of an NIH/3T3 fibroblast bead-loaded as in (a) clearlyshowing diffuse QD labeling. Scale bars are 50 µm.

concentration of extracellular calcium. For this proof-of-principle study, CdSSe/ZnS core/shell QDs (emission max-imum 525 nm, see Fig. S2†) were used as a representativenanomaterial as they are easily detected by conventional fluo-rescence microscopy and can be produced in a range of diam-eters while maintaining homogeneous emission properties.

We first demonstrate that both techniques can deliver QDsto the cytoplasm of NIH/3T3 murine fibroblasts (ATCC CRL-1658), a common cell culture line. Fibroblasts were subjectedto bead- or scratch-loading in the presence of 1-µM QDs andsubsequently visualized by epifluorescence microscopy. Tofirmly establish the intracellular origin of any observed PL,trypan blue was included in the imaging medium to quenchany PL arising from extracellular QDs.31,32 We used trypanblue at a concentration of 40 µg/ml which has a quenchingefficiency of > 90% (determined in vitro, see Fig. S3 and SIfor details†). Despite the inclusion of this quencher, diffusecytoplasmic QD PL was observed in a subset of both bead-

2 | 1–7

(Fig. 2a) and scratch-loaded (Fig. 2c) fibroblasts (also seeFigs. S8, S10†). A close-up image of a single cell more clearlyillustrates the cytoplasmic labeling pattern (Figs. 2e, S8†).The cytoplasmic origin of QD PL is further confirmed by themaintenance of QD-labeling in a subset of treated NIH/3T3 fi-broblasts even after passaging for one generation (Fig. S12†).The incidence of QD-labeled cells was consistent with the pur-ported mechanisms of plasma membrane damage: after bead-loading, labeled cells were clustered together (Fig. 2a); and af-ter scratch-loading, they were present only near wounds in themonolayer (dotted lines, Fig. 2c). These observations showedthat both techniques successfully delivered QDs to the cyto-plasm.

Next we established the generalizability of both tech-niques with respect to cell line. The identical bead- andscratch-loading procedures were performed with WPE1-NB11 prostate epithelial cells (ATCC CRL-2851), a distinctadherent mammalian cell line. Similar to the fibroblasts, asubset of bead- (Fig. 2b) and scratch-loaded (Fig. 2d) ep-ithelial cells exhibited diffuse cytoplasmic QD PL even inthe presence of trypan blue (also see Figs. S9, S11, S13†).However, minor differences were observed between cell lines.In comparison with fibroblasts, epithelial cells exhibited in-creased punctate QD PL (arrow V, Fig. 2b), indicative of QDuptake through endocytosis. If endocytosis occurs during theprocedure, it may be suppressed by performing the techniquesat 4◦C. Alternately, if QDs bind non-specifically to the cellsurface and are not removed during washes, endocytosis mayoccur after the procedure. This could be minimized changingthe passivating ligands used during QD water solubilizationor pre-blocking QDs with serum to prevent non-specific bind-ing. The integrity of the epithelial cell monolayer was alsomaintained better after treatment. These results suggest thatcell-line-dependent properties such as rate of endocytosis andadhesion to the substratum may be a primary source of vari-ability among cell lines.

Epifluorescence observation indicated that the two mechan-ical force delivery methods successfully introduced QDs tomany cells in the population (Figs. S10, S11†), making themamenable to high-throughput studies (as opposed to, e.g. mi-croinjection delivery). Flow cytometry was employed toquantify the extent of QD labeling in cell populations aftertreatment. NIH/3T3 fibroblasts were subjected to bead- orscratch-loading with 1-µM QDs and then immediately ana-lyzed on a flow cytometer for QD PL and viability (Fig. 3).Representative histograms of QD PL from bead- and scratch-loaded fibroblasts are shown in Figs. 3a and b. QD-labeledcells were clearly distinguished from unlabeled cells as a sub-population which exhibited broad, elevated PL (insets). Thisis consistent with epifluorescence observation, where success-ful QD delivery was observed only in a subset of cells (Figs. 2,S10, S11†).

x10

2 3 4

a

log QD Photoluminescence

x10

2 3 4

2 3 4

b

untreated + bead + scratch

2 3 4

2

4

6

8

10

% L

ab

ele

d

with

QD

s

d

cont

rol

+ be

ad

+ sc

ratch

untre

ated

cont

rol

+ be

ad

+ sc

ratch

untre

ated

elec

tropo

rate

d

20

40

60

80

100

% V

iab

ility

c

*

**

ns

Fig. 3 Flow cytometric analysis of NIH/3T3 fibroblasts imme-diately after bead- and scratch-loading with QDs. Representa-tive PL histograms of bead-loaded (solid line, a), and scratch-loaded (solid line, b) fibroblasts. PL histograms of untreatedfibroblasts, which were incubated with QDs for the same dura-tion but not subjected to either treatment, are shaded in gray(dotted line). (insets) The same histograms as in the mainchart with counts multiplied by 10 for clarity. Black arrow indi-cates PL intensity below which 99% of the untreated populationfalls. Cells with PL intensity greater than this value were con-sidered to be successfully labeled by QDs. (c) Cell viability, de-termined by PI staining, of fibroblasts after various treatments.Control population represents cells which were not exposed toQDs at all. (d) Percentage of fibroblasts successfully labeledwith QDs after various treatments. Statistical significance: anasterisk * indicates a p-value < 0.05 for shown comparison;where no specific comparison is shown, p-value < 0.05 for allcomparisons. “ns” indicates that the difference is not significant(p-value > 0.05). See SI† for details on significance tests.

Simultaneous viability staining with propidium iodide(PI) revealed that bead- and scratch-loaded cells maintainedgreater than 95% viability, which was not significantly dif-ferent (p-value >0.15) from either control or untreated cells(Fig. 3c). In comparison, fibroblasts subjected to a typicalelectroporation protocol showed significantly compromisedviability (see SI for protocol†). We also observed that QD-labeled cells could be passaged through at least one generation(see Figs. S12, S13†). This is another indication that neitherthe mechanical delivery nor the QD-labeling adversely affectscell viability.

To complete the evaluation of population-level delivery, wequantified the percent of cells labeled with QDs. As a conser-vative estimate of QD-labeling, cells with PL intensity greater

1–7 | 3

than 99% of the untreated population (black arrow, insets ofFigs. 3a and b) were considered successfully labeled. By thismetric, bead- and scratch-loading delivered QDs to 7 ± 1%and 9±2% of NIH/3T3 fibroblasts, respectively (Fig. 3d).

We next investigated the efficiency of the two deliverymethods. To this end, we designed a flow cytometric assayto simultaneously probe the incidence of plasma membranedamage and QD-labeling. 70-kDa tetramethylrhodamine dex-tran (TMR-Dextran) was used as a standard marker of plasmamembrane damage33 and included with QDs in the loadingsolution during treatment. Two parameter scatter plots ofuntreated, bead-loaded, and scratch-loaded NIH/3T3 fibrob-lasts are shown in Figs. 4a-c. TMR-Dextran labeling revealedthat a larger percent of bead-loaded cells, 18±3%, sustainedmembrane damage when compared with scratch-loaded cells,14±2% (Fig. 4d). Of this damaged population, 36±3% and51±2% of bead- and scratch-loaded cells, respectively, weresuccessfully labeled with QDs (data analysis is described inthe SI, see Figs. S4-S7†). Both of these delivery efficien-cies compare favorably with McNeil and co-workers’ deliveryof fluorescent dextrans28,34 and with QD delivery by a mi-crofluidic device.35 This suggests that these simple mechani-cal stress techniques are largely undiscriminating towards ex-ogenous materials, thereby mediating the delivery of nanopar-ticles nearly as efficiently as that of soft macromolecules. Itis expected that the yield of labeled cells can be proportion-ally increased by increasing the number of cells which incurplasma membrane damage. Scratch-loading was also slightlymore efficient than bead-loading at delivering QDs (∼50%versus ∼40% efficiency), indicating a subtle difference in theirmechanism of action. Thus gains in either throughput or de-livery efficiency may be realized by new delivery strategies,which rely on the same general mechanism but are executedby different mechanical actions.

To further establish the scope, we investigated the efficiencyof delivering a range of QD sizes to NIH/3T3 fibroblasts. Asize panel of QDs with similar emission maxima but differentdiameters (7, 10, and 15 nm) was synthesized by coating thesame CdSSe core with increasing monolayers of ZnS (see Ta-ble S1 and Fig. S1†). After transfer to aqueous buffer, thehydrodynamic diameter of QDs was increased by +18–20 nm(Table S1 and Fig. S1†). The results of this size dependent as-say are summarized in Fig. 4e. Again, TMR-Dextran was in-cluded in the loading solution to serve as a marker of plasmamembrane damage. Both techniques resulted in the quanti-tative delivery of all sizes of QDs. Notably, the efficiencyof scratch-loading exhibited little size dependence ( p-value> 0.08 for all comparisons). In contrast, bead-loading wasfound to be modestly dependent on QD size, with 7-nm QDsshowing significantly ( p-value < 0.05) higher delivery effi-ciency than either 10-nm or 15-nm QDs. This result suggeststhat scratch-loading induces larger membrane disruptions than

untre

ated

+ be

ad

+ sc

ratch 7 10 15 7 10 15

+ bead + scratch

QD (nm):

30

40

50

% E

ffic

ien

cy

20

60

5

10

20

% L

ab

ele

d w

ith

TM

R-D

extr

an

2 3 4 5 2 3 4 5 2 3 4 5

2

3

4

5

log QD Photoluminescence

log

TM

R-D

extr

an

Flu

ore

sce

nce

untreated + bead + scratch

a b c

d e

15

QI QIII

*

*ns

*

*

*

QIV

Fig. 4 Efficiency of QD labeling by bead- and scratch-loading.Representative two parameter scatter plots of untreated (a),bead-loaded (b), and scratch-loaded (c) NIH/3T3 fibroblastswhere TMR-Dextran and QDs were included in the loading so-lution. Dashed lines divide events into quadrants, numbered in(a), which were used to calculate efficiency. Quadrants I andIII designate cells which sustained membrane damage (TMR-Dextran positive); quadrant III designates cells which both sus-tained membrane damage and were labeled with QDs (TMR-Dextran and QD positive). (d) Percent of fibroblasts labeledwith TMR-Dextran after various treatments. (e) Efficiency oflabeling fibroblasts by three sizes of QDs, calculated as thepercent of cells which sustained plasma membrane damagewhich also exhibited successful labeling with QDs. Statisticalsignificance: an asterisk * indicates a p-value < 0.05 for showncomparison; where no specific comparison is shown, p-value< 0.05 for all comparisons. “ns” indicates that comparison isnot significant (p-value > 0.05). See SI† for details on signifi-cance tests.

bead-loading. Regardless of technique, the measured deliveryefficiency of 30−50% for all sizes of QDs indicates that thesedisruptions must be much larger than the maximum hydrody-namic QD diameter of ∼34 nm evaluated here. On this basis,we anticipate that mechanical force based delivery methods, inparticular the simple and accessible bead- and scratch-loadingmethods, will be suitable for the cytoplasmic introduction ofother structurally diverse nanomaterials.

An important goal of cytoplasmic delivery is to targetnanoparticles and nanostructures to organelles or proteins tospecifically image, sense, or manipulate their local environ-ment. Thus as a final evaluation of these two mechanicalstress techniques, we illustrate that QDs introduced by bead-and scratch-loading could be successfully translocated to anintracellular target. Nuclear targeting was chosen as a suit-

4 | 1–7

Hoechst TMR-Dextran QD Merge

NIH

/3T

3+

be

ad

W

NW NW NW NW

W W W

a

b

NW NW NW NW

WW W WWP

E1

-NB

11

+ s

cra

tch

Fig. 5 Nuclear targeting of NLS-QDs after introduction bybead- and scratch-loading. NIH/3T3 fibroblasts bead-loaded(a) and WPE1-NB11 cells scratch-loaded (b) with TMR-Dextran and NLS-QDs. TMR-Dextran (red) serves as a markerof plasma membrane damage. NLS-QDs (green) were translo-cated to the nucleus, based on co-localization with Hoechst33342 (blue), only in wounded cells (e.g. marked by W arrow).In contrast, no nuclear labeling by NLS-QDs is visible in non-wounded cells (e.g. marked by NW arrow). Scale bars are 25µm.

able proof-of-concept because it provided clear evidence ofsuccessful targeting in optical imaging based assays. Pep-tides derived from the nuclear localization sequence (NLS) ofthe SV-40 large T-antigen have previously been used to tar-get fluorophores,36 proteins,37 and QDs19,38 to the nucleus ofmammalian cells. We made use of the extensively reported39

method of histidine-tag coordination to the ZnS surface ofQDs to produce robust, targeted NLS-QD conjugates. Fig. 5ashows representative epifluorescence images of NIH/3T3 fi-broblasts three hours after bead-loading with NLS-QDs andTMR-Dextran, which again served as a marker of plasmamembrane disruption (also see Fig. S14†). Nuclear localiza-tion of NLS-QDs (green), confirmed by co-localization withcell-permeant DNA stain Hoechst 33342 (blue), was observedonly in membrane-disrupted cells (red). Similar results wereobtained when the same assay was repeated with scratch-loading in WPE1-NB11 cells (Fig. 5b, S15†). The observa-tion of nuclear localization only in those cells also labeled byTMR-Dextran confirms that plasma membrane tearing is nec-essary for targeting to occur. Importantly, this experiment es-tablishes that the functional moieties grafted on QD surface re-main accessible for biochemical recognition once introducedto the cytoplasm. Thus we anticipate that both bead- andscratch-loading will be useful for implementing any numberof orthogonal strategies for targeting nanoparticles to specificintracellular proteins or organelles.

The development of a myriad of functional nanomaterialsfor sensing and manipulating chemical systems has excitingimplications for biology and medicine. However, the deliveryof arbitrary nanomaterials to living systems remains a persis-

tent obstacle to the implementation of these novel technolo-gies. Although many efficient solutions to the problem ofnanomaterial delivery have been proposed, they typically suf-fer from steep trade-offs between generality and accessibil-ity. Thus there is a need for the community to have a sim-ple and widely deployable method that can effectively delivernanoparticles and nanostructures into the cytoplasm of livingcells. In this report, we have identified that a mechanicalforce based delivery modality is general because it does notrely on cellular mechanism nor does it require special surfacechemistry for the delivery. For the latter accessible require-ment, inspired by the pioneering works of McNeil et al., wehave been able to expand the scope of two of their deliverymethods—which are also based on mechanical forces—andgeneralize them to delivering nanoparticles to live-cell cyto-plasm. Together with other delivery methods based on me-chanical forces,19,25,26 the present work allowed us to concep-tualize “mechanodelivery,” which emerges as a general nano-material delivery modality. While we have demonstrated andcharacterized in detail two execution modes—the bead- andthe scratch-loading methods—as simple, robust, and broadlyaccessible, we anticipate that additional innovative executionmodes which exploit the mechanodelivery concept will proveuseful for introducing a structurally diverse array of functionalnanomaterials to the cytoplasm.

Experimental

Cell culture

All cell lines were maintained at 5% CO2 and 37◦C.NIH/3T3 murine fibroblasts (ATCC CRL-1658) were culturedin DMEM (Gibco) supplemented with 10% bovine serum andused at passages 20–40. WPE1-NB11 prostate epithelial cells(ATCC CRL-2851) were cultured in Keratinocyte serum freemedium (Gibco) supplemented with 0.05 mg/ml BPE and 5ng/ml EGF and used at passages 14–25. Cells were seeded on25-mm glass coverslips (Warner Instruments) and cultured for2–3 days (70–80% confluence) before treatment. For fibrob-lasts only, coverslips were coated with 10 µg/ml 30–70 kDapoly-D-lysine (Sigma) before plating cells.

Synthesis and passivation of CdSSe/ZnS core/shell QDs

CdSSe cores were synthesized following a previously reportedprocedure40. Monolayers (MLs) of ZnS were added to CdSSecores following the SILAR procedure41, were 0.8 ML of ZnSwas added each cycle40. A size panel of QDs with nom-inal diameters of 7, 10, and 15 nm was produced by coat-ing CdSSe cores with 4, 8, and 12 MLs of ZnS, respectively.The actual QD diameters were measured in toluene with dy-namic light scattering (90-Plus, Brookhaven Instruments) to

1–7 | 5

be 7.2±0.8 nm, 10.5±0.2 nm, and 14.7±0.1 nm (see Table S1and Fig. S1†).

QDs were transferred to the aqueous phase by passi-vating with dithiol methoxy poly(ethylene glycol) ligands(lipoamide-dPEG12, Quantabiodesign). Detailed descriptionof phase transfer and passivation procedure is supplied inthe SI†. The diameters of PEG-coated 7, 10, and 15nm QDs were measured again with DLS and found to be27.2±1.0, 28.9±3.5, and 33.9±0.6, respectively (see Table S1and Fig. S1†). For nuclear targeting, QDs were incubated withHis-NLS peptide (see SI for sequence†)19 at a peptide:QD ra-tio of 100:1 for 30–45 minutes and used immediately for load-ing.42

Bead- and scratch-loading procedures

Cells grown on coverslips were washed with 1% Pluronic F-68(Gibco) in Dulbecco’s phosphate buffered saline without cal-cium and magnesium (DPBS- -, Gibco). All liquid was aspi-rated and 50-µl loading solution containing 1-µM QDs and/or1-mg/ml TMR-Dextran (Molecular probes) was applied tothe surface. Unless otherwise noted, 10-nm QDs were used.Bead- or scratch-loading was then performed as described byMcNeil and co-workers28,29 using 200-300 µm acid-washedglass beads (Sigma) or a 30-gauge needle (BD), respectively.The duration of exposure to loading solution was held con-stant at 45-seconds for all procedures. After treatment, cellswere washed with warmed DPBS with calcium and magne-sium (DPBS++, Gibco). Treated cells were then returned toculture in complete medium.

Microscopy

Coverslips with treated cells were imaged in a live cell imag-ing chamber (Warner Instruments) on an inverted epifluores-cence microscope (AxioObserver, Zeiss) with either a 20x0.5NA air objective or a 40x 1.3NA oil immersion objective(EC Plan Neofluar, Zeiss). DPBS++ was used as an imag-ing medium; where noted, trypan blue (Gibco) was includedat a final concentration of 40-µg/ml. For nuclear staining,cells were incubated with 10-µg/ml Hoechst 33342 (MolecularProbes) for 10 minutes. For epifluorescence, a mercury lamp(Series 120Q, XCite) was used as an excitation source. Im-ages were acquired with an air-cooled CCD camera (Orca-R2,Hamamatsu) that was controlled by a commercial softwarepackage (HCImageLive). False coloring, overlays, and imageadjustment were carried out with Adobe Photoshop. The ex-citation (EX), dichroic (DC) and emission (EM) filters used toimage fluorophores or QDs were as follows. QDs: EX 470/40,DC 495, EM 525/50 (Filter set 38-HE, Zeiss); TMR-Dextran:EX 546/10, DC 565, EM 620/40 (Chroma); Hoechst 33342:EX 436/25, DC 455, EM 480/45 (Filter set 47, Zeiss).

Flow cytometry

Cells were treated on coverslips, recovered in completemedium for approximately 10 minutes, and collected bytrypsinization into DPBS- - containing 2% fetal bovine serum(FBS, Gibco). Samples were held on ice until analysis ona BD LSR-II cytometer (BD Biosciences). To assay viabil-ity (Fig. S4†), propidum iodide (PI, Molecular Probes) wasadded to a final concentration of 1-µg/ml immediately beforeanalysis. At minimum 10,000 events, considered to be cellsbased on forward and side scatter (see Fig S3†) were recordedfor each sample. The channels, laser excitation wavelengths(EX) and emission filters (EM) used to collect QD or fluo-rophore emission were as follows. QDs: FITC-A channel,EX 488, EM 525/50; PI: PI-A channel, EX 488, EM 610/20;TMR-Dextran: DsRed channel, EX 561, EM 585/42. Datawas analyzed using a commercial software package (FlowJo);a detailed description of this analysis is described in the SI†.

Acknowledgement

The authors thank Christina DeCoste and John Grady at thePrinceton University flow cytometry facility for assistancewith cytometer operation and training. This work was sup-ported by the Eric and Wendy Schmidt Transformative Tech-nology Fund.

References1 A. Taylor, K. M. Wilson, P. Murray, D. G. Fernig and R. Levy, Chem Soc

Rev, 2012, 41, 2707–2017.2 W. Baschong, J. M. Lucocq and J. Roth, Histochemistry, 1985, 83, 409–

411.3 M. Bruchez, M. Moronne, P. Gin, S. Weiss and A. P. Alivasatos, Science,

1998, 281, 2013–2016.4 C.-C. Fu, H.-Y. Lee, K. Chen, T.-S. Lim, H.-Y. Wu, P.-K. Lin, P.-K. Wei,

P.-H. Tsao, H.-C. Chang and W. F, Proc. Natl. Acad. Sci. U.S.A., 2007,104, 727–732.

5 F. Pinaud, S. Clarke, A. Sittner and M. Dahan, Nat. Methods, 2010, 7,275–285.

6 J. D. Krooswyk, C. M. Tyrakowski and P. T. Snee, J. Phys. Chem. C,2010, 114, 21348–21352.

7 I. L. Medintz, M. H. Stewart, S. A. Trammell, K. Susumu, J. B. Dele-hanty, B. C. Mei, J. S. Melinger, J. B. Blanco-Canosa, P. E. Dawson andH. Mattoussi, Nat. Mater., 2010, 9, 676–684.

8 S. Li, K. Zhang, J.-M. Yang, L. Lin and H. Yang, Nano Lett., 2007, 7,3102–3105.

9 J.-M. Yang, H. Yang and L. Lin, ACS Nano, 2011, 5, 5067–5071.10 G. Kucsko, P. C. Maurer, N. Y. Yao, M. Kubo, H. J. Noh, P. K. Lo, H. Park

and M. D. Lukin, Nature, 2013, 500, 54–58.11 I. R. Nabiev, H. Morjani and M. Manfait, Eur. Biophys. J., 1991, 19, 311–

316.12 N. K. Mal, M. Fujiwara and Y. Tanaka, Nature, 2003, 421, 350–353.13 T. R. Kuo, V. A. Hovhannisyan, Y. C. Chao, S. L. Chao, S. J. Chiang,

S. J. Lin, C. Y. Dong and C. C. Chen, J. Am. Chem. Soc., 2010, 132,14163–14171.

6 | 1–7

14 S. Biswas, K. Kinbara, T. Niwa, H. Taguchi, N. Ishii, S. Watanabe,K. Miyata, K. Kataoka and T. Aida, Nat. Chem., 2013, 5, 613–620.

15 H. de Puig, A. C. Rius, D. Flemister, S. H. Baxamusa and K. Hamad-Schifferli, PLoS One, 2013, 8, e68511.

16 F. Scherer, M. Anton, U. Schillinger, J. Henke, C. Bergemann, A. Krugerand B. Gansbacher, Gene Therapy, 2002, 9, 102–109.

17 C. Hoffmann, E. Mazari, S. Lallet, R. L. Borgne, V. Marchi, C. Gosse andZ. Gueroui, Nat. Nanotech., 2013, 8, 199–205.

18 D. Luo and W. M. Saltzman, Nat. Biotechnol., 2000, 18, 33–37.19 A. M. Derfus, W. C. W. Chan and S. N. Bhatia, Adv. Mater., 2004, 16,

961–966.20 L. Y. T. Chou, K. Ming and W. C. W. Chan, Chem. Soc. Rev., 2011, 40,

233–245.21 C. M. Tyrakowski and P. T. Snee, Phys. Chem. Chem. Phys., 2014, 16,

837–855.22 J. B. Delehanty, J. B. Blanco-Canosa, C. E. Bradburne, K. Susumu, M. H.

Stewart, D. E. Prasuhn, P. E. Dawson and I. L. Medintz, Chem. Commun.,2013, 49, 7878–7880.

23 B. Y. S. Kim, W. Jiang, J. Oreopolulos, C. M. Yip, J. T. Rutka andW. C. W. Chan, Nano Lett., 2008, 8, 3887–3892.

24 A. R. Bayles, H. S. Chahal, D. S. Chahal, C. P. Goldbeck, B. E. Cohenand B. A. Helms, Nano Lett., 2010, 10, 4086–4092.

25 A. K. Shalek, J. T. Robinson, E. S. Karp, J. S. Lee, D.-R. Ahn, M.-H.Yoon, A. Sutton, M. Jorgolli, R. S. Gertner, T. S. Gujral, G. MacBeath,E. G. Yang and H. Park, Proc. Natl. Acad. Sci. U.S.A., 2010, 107, 1870–1875.

26 A. Sharei, J. Zoldan, A. Adamo, W. Y. Sim, N. Cho, E. Jackson, S. Mao,S. Schneider, M.-J. Han, A. Lytton-Jean, P. A. Basto, S. Jhunhuhwala,J. Lee, D. A. Heller, J. W. Kang, G. C. Hartoularos, K.-S. Kim, D. G.Anderson, R. Langer and K. F. Jensen, Proc. Natl. Acad. Sci. U.S.A., 2013,

110, 2082–2087.27 P. McNeil, Meth. Cell. Biol., 1989, 29, 153–173.28 P. McNeil and E. Warder, J. Cell Sci., 1987, 88, 669–78.29 J. A. Swanson and P. L. McNeil, Science, 1987, 238, 548–550.30 R. Steinhardt, G. Bi and J. M. Alderton, Science, 1994, 263, 390–393.31 A. E. Jablonski, T. Kawakami, A. Y. Ting and C. K. Payne, J. Phys. Chem.

Lett., 2010, 1, 1312–1315.32 M. Howarth, W. Liu, S. Puthenveetil, Y. Zhen, L. F. Marshall, M. M.

Schmidt, K. D. Wittrup, M. G. Bawendi and A. Y. Ting, Nat. Methods,2008, 5, 2007–2009.

33 K. Miyake, P. McNeil, K. Suzuki, R. Tsunoda and N. Sugai, J. Cell Sci.,2001, 114, 3487–3494.

34 P. McNeil, R. F. Murphy, F. Lanni and D. L. Taylor, J. Cell Biol., 1984,98, 1556–64.

35 J. Lee, A. Sharei, W. Y. Sim, A. Adamo, R. Langer, K. F. Jensen andM. G. Bawendi, Nano Lett., 2012, 12, 6322–6327.

36 E. H. W. Pap, T. B. Dansen, R. van Summeren and K. W. A. Wirtz, Exp.Cell Res., 2001, 265, 288–293.

37 D. S. Goldfarb, J. Gariepy, G. Schoolnik and R. D. Kornberg, Nature,1986, 322, 641–644.

38 F. Chen and D. Gerion, Nano Lett., 2004, 4, 1827–1832.39 I. Medintz, A. R. Clapp, H. Mattoussi, E. R. Goldman, B. Fisher and J. M.

Mauro, Nat. Mater., 2003, 2, 630–638.40 C.-H. Hsia, A. Wuttig and H. Yang, ACS Nano, 2011, 5, 9511–9522.41 J. J. Li, Y. A. Wang, W. Guo, J. C. Keay, T. D. Mishima, M. B. Johnson

and X. Peng, J. Am. Chem. Soc., 2003, 125, 12567–12575.42 J. B. Delehanty, C. E. Bradburne, K. Boeneman, K. Susumu, D. Farrell,

B. C. Mei, J. B. Blanco-Canosa, G. Dawson, P. E. Dawson, H. Mattoussiand I. L. Medintz, Integr. Biol., 2010, 2, 265–277.

1–7 | 7