Optical coding of mammalian cells using semiconductor quantum dots

ANALYTICAL
BIOCHEMISTRY
Analytical Biochemistry 327 (2004) 200–208
www.elsevier.com/locate/yabio

Optical coding of mammalian cells using semiconductor
quantum dots
Larry C. Mattheakis,Ô Jennifer M. Dias, Yun-Jung Choi, Jing Gong,
Marcel P. Bruchez, Jianquan Liu, and Eugene Wang
Quantum Dot Corp., 26118 Research Road, Hayward, CA 94545, USA
Received 25 September 2003

Abstract
Cell-based assays are widely used to screen compounds and study complex phenotypes. Few methods exist, however, for multiplexing cellular assays or labeling individual cells in a mixed cell population. We developed a generic encoding method for cells that is based
on peptide-mediated delivery of quantum dots (QDs) into live cells. The QDs are nontoxic and photostable and can be imaged using
conventional Xuorescence microscopy or Xow cytometry systems. We created unique Xuorescent codes for a variety of mammalian cell
types and show that our encoding method has the potential to create 1100 codes. We demonstrate that QD cell codes are compatible
with most types of compound screening assays including immunostaining, competition binding, reporter gene, receptor internalization,
and intracellular calcium release. A multiplexed calcium assay for G-protein-coupled receptors using QDs is demonstrated. The ability
to spectrally encode individual cells with unique Xuorescent barcodes should open new opportunities in multiplexed assay development
and greatly facilitate the study of cell/cell interactions and other complex phenotypes in mixed cell populations.
 2004 Elsevier Inc. All rights reserved.
Keywords: Quantum dots; Multiplexed cell-based assays; Spectral encoding; Receptor; Nanocrystal

The growing size of compound libraries and therapeutic targets has driven the need for new screening technologies. The desire to develop new methods for
massively parallel analyses has led to the development of
microarray chips [1–5] and encoded microsphere beads
[6–11] for use as biosensors and for studying nucleic
acids and proteins. These methods have been useful for
studying biochemical interactions, but there has been
limited progress to extend these approaches to cell-based

screening. Cell-based assays are widely used to screen the
activities of compounds against important membrane
receptor targets or to provide important preclinical data
on a compound’s toxicity or bioavailability.
To multiplex cell-based assays, it is possible to use
positional cell arrays, but these systems require sophisticated robotic systems or unique substrate surfaces that
are cell-type speciWc. Cell patterning via surface modiW-

cation of the substrate can be accomplished by chemical,
photochemical, or lithographic methods [12–16]. Microfabrication of nanowells on a membrane surface has also
been used to construct cell microarrays [15]. An alternative approach, transfected cell microarray, is based on
culturing mammalian cells on glass slides printed with
deWned cDNAs [17]. The cells take up the DNA and
create deWned locations of transfected cells on the slide
surface.
To create a more versatile multiplexing strategy for
cell-based assays, it would be desirable to encode individual cells with unique identiWer barcodes. Such a system could then be used for a variety of cell types and
would not require that cells adhere to an array surface.
Encoded cells would also be compatible with standard
single cell analysis platforms such as microscopy or a
Xuorescence-activated cell sorter (FACS).1

Ô
Corresponding author. Present address: Cytokinetics, Inc., 280
East Grand Ave., South San Francisco, CA 94080, USA; Fax: 1-650624-3010.
E-mail address: (L.C. Mattheakis).

1
Abbreviations used: FACS, Xuorescence-activated cell sorter; QD,
quantum dot; CHO, Chinese hamster ovary; GPCR, G-proteincoupled receptor, HA, hemagglutinin; PBS, phosphate-buVered saline.

0003-2697/$ - see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.ab.2004.01.031


L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

Here we describe a cell encoding technology based on
quantum dot (QD) nanocrystals. QDs are nanometersized crystals of semiconductor material, typically 2–
6 nm in diameter. Although they are chemically identical
to bulk semiconductor material, they exhibit optical
properties that are highly dependent on their size [18].
QDs can be excited to emit light in a manner analogous
to that of organic Xuorophores. Unlike organic dyes,
however, their broad absorption spectrum allows all colors to be excited with a single wavelength of light and
they do not bleach signiWcantly [11,19,20].
Recent advances in QD chemistry have made it possible to transfer quantum dots into aqueous buVers and to
modify the surface so that biological aYnity molecules
such as antibodies and nucleic acid probes can be
attached and used as direct labels to detect biological
markers in various applications [20–27].
QDs have also been used to label live cells. The earliest example showed that a transferrin–QD conjugate is
transported into live HeLa cells by receptor-mediated
endocytosis [23]. More recent work has shown that QDs
encapsulated in phospholipid micelles can be injected
into Xenopus embryos and used to trace cell lineage [26].
Finally, multicolor QDs were shown to be useful for
visualizing the eVects of diVerent starvation times on
aggregation of Dictyostelium discoideum [28].
Here we describe a QD system for encoding, imaging,

and decoding single cells for multiplexing and other
assay applications (Fig. 1). We developed a generic
method to deliver multicolor QDs into live mammalian
cells and show that QDs are compatible with a variety of
important drug-screening cell-based assays. Each cell
type is encoded separately with a unique and spectrally
resolvable QD code. The encoded cells are mixed and

201

aliquots of the mixture are deposited into the wells of an
assay plate. We used a microscope-based imaging system
to identify and decode individual cells and show examples of using this encoding scheme to multiplex cellbased assays.

Materials and methods
Preparation of water-soluble quantum dots
Organic-soluble, CdSe/ZnS core-shell nanocrystals
[19,29] were isolated from hexanes and ligand solution
with an equal volume of methanol, rinsed with methanol, and redispersed in CHCl3. These materials were
mixed with neutralized amphiphilic polymer (40% octylamine-modiWed polyacrylic acid, 2000 units/QD) in
CHCl3, and the solvent was evaporated. The dry Wlm
was redispersed in water and puriWed from excess
polymer by gel Wltration. The surface coating was crosslinked further with 1-ethyl-3-(3-dimethylamino propyl)
carbodimide-mediated coupling to lysine (or polyethylene glycol-lysine). These materials were then puriWed by
gel Wltration and ion exchange spun columns in the
presence of 10 mM borate buVer, pH 8.2.
Delivery of QDs into cells using Pep-1
CHO-KI cells (ATCC) were incubated at 37 °C, 5%
CO2 in growth medium consisting of Dulbecco’s modiWed Eagle’s medium/nutrient mixture F12 (DMEM/F12)
containing 5% serum. Cells were seeded in 35-mm-diameter wells and grown to a density of 3 £ 105 cells per well.

The Pep-1 peptide is available commercially as Chariot

Fig. 1. Schematic illustration of a multiplex cellular screening assay using QDs. DiVerent cell lines or a common host cell line expressing diVerent
receptors are encoded separately with QD cell codes. The cells are mixed and aliquots from the master mix are deposited into the wells of a clear-bottomed assay plate. Compounds are added to the wells of the assay plate and the plate is imaged using an inverted-microscope-based imaging system.
Alternatively, the cells are removed from the assay plate and analyzed by FACS.


202

L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

from Active Motif (Carlsbad, CA). To form a complex
between QDs and Pep-1, QDs were diluted in phosphate-buVered saline (PBS) (pH 7.2) to a Wnal volume of
0.1 ml and added to an equivalent volume of Pep-1
diluted in water. The complex was incubated for 30 min
at room temperature. The cell culture medium was
removed, and the 0.2-ml complex was added Wrst to cells,
followed by 0.4 ml of DMEM/F12. The cells were incubated for 1 h and 1 ml of DMEM/F12 containing 5%
serum was added. The cells were incubated for an additional 2 h and either lifted for analysis or incubated overnight as described.
Cloning and expression of epitope-tagged GPCRs
The cDNAs for the
2-adrenergic, serotonin 2A, and
serotonin 2B receptors were cloned from human brain
mRNA or cDNA libraries (BD Biosciences Clontech,
Palo Alto, CA). The sequence for the nine-amino acid
hemagglutinin (HA) epitope was inserted at the 50 end of
the coding sequence and the genes were cloned into the
eukaryotic expression vector pcDNA3.1 (Invitrogen,
Carlsbad, CA). CHO cells were transfected and selected
in the presence of 800 g/ml G418. Clones expressing
receptor were isolated by FACS using a Xuorescent antibody directed against the HA sequence.

Receptor internalization assay
CHO cells expressing the HA-tagged
2 adrenergic
receptor were preincubated with a mouse monoclonal
anti-HA antibody (CRP Inc., Denver, PA) for 30 min.
Cells were incubated in the presence or absence of agonist (10 M isoproterenol), Wxed with 3.7% formaldehyde in PBS, and permeabilized with 0.1% Triton X-100
in Blotto buVer (3% dry milk, 25 mM Tris–HCl, pH 7.4,
137 mM NaCl, 3 mM KCl, 1 mM CaCl2). The samples
were incubated with Xuorescein isothiocyanate-conjugated goat anti-mouse IgG (Jackson ImmunoResearch,
West Grove, PA) for 30 min and imaged by Xuorescence
microscopy.
Multiplex calcium assay and imaging
Cells were seeded into the wells of a clear-bottomed
96-well plate and loaded with a calcium dye using the
FlexStation calcium assay kit (Molecular Devices,
Sunnyvale, CA). The cells were imaged using the Discovery-1 microscope system (Universal Imaging Corp.,
Downingtown, PA). The calcium Xuorescence image of
cells was acquired before and approximately 7 s after
addition of carbachol (30 M Wnal concentration). To
determine the QD codes, the same Weld of view was
imaged with 15-nm-bandwidth emission Wlters centered
at 560, 575, 590, 605, 620, and 635 nm. The calcium
Xuorescence fold induction was measured using the

Metamorph V version 4.5 software package (Universal
Imaging Corp.).

Results
Transfecting QDs into live cells
To deliver quantum dots into live cells, we explored
the use of peptide translocation domains, cationic lipids,

and polymeric micelles. All of these methods resulted in
the internalization of quantum dots, but the percentage
of labeled cells and eVects on cell viability varied
depending on the method and QD surface properties.
These results demonstrate that a variety of methods exist
for labeling cells with QDs.
We found that the protein translocation domain
Pep-1 is an eYcient and convenient carrier for delivering
QDs into cells. Pep-1 is a synthetic 21-residue amphipathic peptide composed of three functional sequences:
a hydrophobic tryptophan-rich sequence, a spacer
sequence, and a hydrophilic lysine-rich sequence from
the nuclear localization sequence of simian virus 40 large
T antigen [30]. The Pep-1 peptide is nontoxic and has
been shown to deliver a variety of peptides and proteins
into cells [30]. The hydrophobic sequence binds to proteins or QDs noncovalently, and the lysine-rich sequence
functions to penetrate cells and deliver the bound complex intracellularly.
We incubated a Wxed concentration of Pep-1 with
increasing concentrations of QDs and tested each complex for its ability to penetrate Chinese hamster ovary
cells as measured by a FACS. Fig. 2A shows that the
mean cell Xuorescence reaches a maximum at 5 nM for a
560-nm-emitting QD but then decreases to near background levels at 10 nM. A similar bell-shaped curve was
observed for other QD materials, but the absolute optimum QD concentration varied depending on the speciWc
QD material used in the experiment (data not shown).
Fig. 2A also indicates that QD penetration into cells is
dependent on the carrier Pep-1.
To study this concentration dependence in more
detail, we Wxed the 560-nm QD concentration at 10 nM
and measured the eVect of increasing peptide concentration. Fig. 2B shows that doubling the QD concentration
required a concomitant twofold increase in peptide concentration to obtain the maximum increase in cellular
Xuorescence. Increasing the peptide concentration

beyond 60 M did not increase QD internalization.
Together, these results suggest that multiple copies of
carrier peptide are associated per QD and that an
optimum number is required for eYcient delivery into
cells. This conclusion is supported by a previous study
that showed, for green Xuorescent protein and
-galactosidase, that a molar ratio of Pep-1:protein of 40:1
was optimal for delivery into Cos-7 or HS-68 cells [30].


L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

203

Fig. 2. EVect of varying QD (560 nm) or Pep-1 concentration on QD delivery into cells. (A) QDs were incubated in the presence (squares) or absence
(triangles) of 30 M Pep-1. (B) Pep-1 was incubated in the presence (squares) or absence (triangles) of 10 nM QDs. The complexes were added to
CHO cells as described under Materials and methods. The cells were incubated overnight and lifted, and the Xuorescence intensity was measured by
FACS. The mean Xuorescence units are the average of 10,000 events.

Our data estimates an optimized Pep1:QD ratio, for
CHO cells, of approximately 6000:1 for the 560-nm QD
material shown in Fig. 2.
We also measured the percentage of QD-labeled cells
by Xuorescence microscopy. Nearly 100% of CHO cells
contain QDs under these optimized conditions. The QDs
are contained within vesicles, but the number of QDcontaining vesicles and the total Xuorescence intensity
varies among individual cells. Confocal microscopy conWrmed that the QDs are intracellular and that the vesicles are distinct from the mitochondria and endoplasmic
reticulum compartments (data not shown).
The Pep-1 transfection method for QDs is not limited
to CHO cells. We used the same procedure to introduce
QDs into a variety of mammalian cell types, but the QD

concentration had to be optimized for each cell type. We
demonstrated QD transfection of HEK293, NIH3T3,
SKBR3, and the primary cell line HUVEC and found
the percentage of transfected cells to range from 80 to
95% under optimized conditions.
We also tracked the stability of QDs inside CHO cells
by comparing the growth rates of cells transfected with
QDs to nontransfected cells over a 6-day period. The
growth rates were similar, and the percentage of labeled
cells decreased in concordance with the number of cell
doublings.
To measure QD toxicity, we used 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT
reagent) to measure cell survival after QD transfection
[31]. Although some QD materials were toxic in this
assay, we found that a combination of anionic, cationic,
and size exclusion puriWcation of QDs after stabilization
could reduce toxicity to near undetectable levels at the
concentrations used for transfection.
Development of cell codes
To detect QDs inside cells, it is possible to use a
FACS or microscope to measure the intensity and wavelengths of the codes. We chose microscopy because it is
more amenable to multicolor imaging, and a greater

variety of cell-based assays are compatible with microscope-based imaging systems compared to Xow-based
systems. Our imaging system is a standard epiXuorescence microscope equipped with a single 488-nm argon
laser and two emission Wlter wheels containing 18 diVerent Wlters. The 10-nm band pass Wlters are evenly spaced
to cover the visible region from 510 to 680 nm.
The Wlters are used to determine the Xuorescence
emission spectra of individual cells. We devised a decoding algorithm based on comparing the pattern of an
unknown cell’s spectrum to stored reference spectra. The

algorithm is based on the Pearson coeYcient, which
describes the strength of the association between the
observed values and the predicted reference values [32].
A coeYcient of 0.95 was used as a cutoV to match an
observed cell’s spectrum to that of the reference.
To test our imaging and decoding system for resolving codes, we incubated the Pep-1 peptide with Wve single-color or Wve dual-color QD combinations and added
the complexes to CHO cells. The cells were incubated
overnight and the spectra of individual cells determined
(Fig. 3). Although the absolute Xuorescence intensity of
individual cells can vary greatly, the curves are quite similar when the area under the curve of each cell’s spectrum
is normalized relative to that of the brightest cell. The
similarity of the normalized curves for the dual color
codes indicates that cells take up diVerent QD colors at
nearly the same ratio for a variety of QD combinations.
We also found that it was possible to vary the concentration ratios of QD colors and obtain additional codes.
For example, the combination of QD colors 582 and
630 nm at 2:1 or 1:1 molar ratios yielded two diVerent
codes as determined by our decoding algorithm (Fig. 3).
To test multiplexing, we set up a mock Wve-plex assay.
Five separate cultures of CHO cells were encoded with
566-, 582-, 608-, 630-, or 647-nm QDs. The cells were
lifted after the encoding step and mixed, and aliquots
were deposited into the wells of a 96-well assay plate for
imaging the next day (Fig. 4A). Fig. 4B shows that the
mixed population of cells could be segregated into Wve
distinct QD codes. These results demonstrate that we


204

L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

Fig. 3. Fluorescence emission spectra of 10 QD cell codes. CHO cells were encoded as described under Materials and methods and Wxed, and the normalized Xuorescence spectra (510–680 nm) were determined from complexes consisting of 40 M Pep-1 and the indicated QD colors (5 nM each). For
the 582- and 630-nm codes, the 2:1 and 1:1 molar ratios for the 582- and 630-nm QDs were 5 nM:2.5 nM and 2.5 nM:2.5 nM, respectively. Shown for
each code are the normalized spectra of approximately 25 cells. Each line is the normalized spectrum of a single cell.

Fig. 4. Imaging and decoding of a mixed cell population. (A) CHO cells were encoded with 566-, 582-, 608-, 630-, or 647-nm QDs, mixed, and transferred to an assay plate for imaging the next day. Cell nuclei were stained with Hoechst 33342. Shown is an image captured using a Nikon D1 color
digital camera. Filter sets for excitation and emission were 390 § 100 nm and 490 nm long pass, respectively. (B) Normalized emission spectra
of approximately 60 cells from the mixed cell population. Each line is the normalized spectrum of a single cell.

can encode, image, and decode mixed cell populations in
the absence of an assay.
Encoded cellular assays
To use QDs for encoding cellular assays, the internalized QDs must not interfere with normal cell physiology
such as signal transduction, receptor traYcking, and
membrane function. We tested the eVects of QDs in a
variety of cell-based assays including immunostaining,
receptor binding, reporter gene expression, receptor
internalization, and intracellular release of calcium.
To test the eVect of QDs in an immunostaining application, we encoded CHO cells with 530-nm QDs. The
cells were Wxed and stained using a Xuorescent antibody
directed against tubulin (Fig. 5A). Control experiments
showed that the staining intensity was equivalent in the
presence or absence of QDs (data not shown).
We also tested the eVect of QDs on binding of a Xuorescent ligand to a cell surface receptor. CHO cells

expressing the
2-adrenergic receptor were encoded with
530-nm QDs and prepared for a competition-binding
assay. The Xuorescent ligand CGP-12177 bound to cells
expressing the receptor (Fig. 5B), although the intensity

of CGP-12177 Xuorescence among individual cells varied, possibly due to diVerences in receptor expression or
quenching of CGP-12177 Xuorescence by the QDs. Binding was blocked in the presence of excess unlabeled
CGP-12177 (Fig. 5C). Therefore, these results indicate
that internalized QD codes do not interfere with binding
of ligands or antibodies to intracellular and membrane
surface targets.
The presence of intracellular vesicles containing QDs
could potentially aVect protein traYcking. To test this,
we measured the agonist-induced internalization of the

2 adrenergic receptor. CHO cells encoded with 608-nm
QDs and expressing an epitope-tagged version of the
2
receptor were incubated in the absence (Fig. 5D) or
presence (Fig. 5E) of the agonist isoproterenol. Fig. 5E
shows that isoproterenol causes the
2 receptor to trans-


L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

205

Fig. 5. Fluorescent microscopy images of encoded cells in various cellular assays. (A) Immunostaining of tubulin in CHO cells. Cells were encoded
with 530-nm QDs and Wxed for immunostaining using rabbit antitubulin IgG fraction (Sigma–Aldrich, St. Louis, MO), biotinylated goat anti-rabbit
IgG (Vector Laboratories, Burlingame, CA), and streptavidin-conjugated Cy3 (Amersham Bioscience, Piscataway, NJ). Shown is a composite of the
Cy3 and QD images. (B and C) Binding of CGP-12177 to CHO cells expressing the
2-adrenergic receptor. CHO cells were encoded with 530-nm
QDs and incubated with 250 nM BODIPY TMR (§) CGP-12177 (Molecular Probes, Eugene, OR) in the absence (B) or presence (C) of 1 M
unlabeled CGP-12177 (Sigma–Aldrich). Shown are composites of the QD and BODIPY images. Binding was measured as total pixel intensity
of CGP-12177 Xuorescence. (D and E) Agonist-induced internalization of the
2-adrenergic receptor. CHO cells expressing HA-tagged
2-adrenergic
receptor were encoded with 608-nm QDs and incubated in the absence (D) or presence (E) of 10 M isoproterenol. The cells were assayed for receptor
internalization as described under Materials and methods. Shown is a composite of the FITC and QD images.

locate from the cell surface to intracellular vesicles of
the encoded cells. To determine the dose response of
translocation, we measured the intensity of intracellular
Xuorescence and found the EC50 values to be similar in
encoded and unencoded cells (data not shown). Thus,
QDs contained within vesicles appear to be relatively
inert and do not interfere with traYcking events within
the cell.
To test the eVects of QDs on signal transduction, we
used a CHO cell line expressing the
2-adrenergic receptor and carrying a luciferase reporter gene under the
transcriptional control of a cyclic adenosine monophosphate (cAMP) responsive promoter. Stimulation of the


2 receptor with the agonist isoproterenol results in elevation of cAMP, which activates transcription of the
luciferase reporter. We found the dose responses in
unencoded cells and encoded cells to be similar and the
EC50 values were approximately 1 nM (Fig. 6).
Finally, to demonstrate the use of QDs in a multiplexing experiment, we encoded three diVerent GPCR cell
lines and simultaneously measured the agonist-induced
eVects on intracellular calcium levels. Cell lines expressing the muscarinic M1, serotonin 2A, and serotonin 2B
receptors were encoded with 605-, 582-, and 630-nm
QDs, respectively. The cell lines were mixed, and aliquots
of the mixture were transferred into the wells of an assay


206

L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

Fig. 6. EVect of QDs on a luciferase dose response assay. Approximately 1 £ 105 of control or 530-nm QD-encoded CHO cells expressing the
2-adrenergic receptor and renilla luciferase reporter gene were
seeded into each well of a 96-well assay plate. Cells were incubated

overnight in DMEM/F12 medium containing 10% serum. The cells
were washed and incubated 24 h in DMEM/F12 medium lacking
serum or phenol red. Cells were incubated with isoproterenol at the
indicated concentrations for 4 h, washed, and assayed for luciferase
activity using the renilla luciferase assay system (Promega Corp., Madison, WI) and SpectraFluor Plus instrument (Tecan, Research Triangle
Park, NC). Shown are the dose responses of encoded (dashed line) and
control (solid line) cells. Values are the average of three measurements.

plate. The next day, the cells were loaded with a Xuorescent calcium indicator and incubated in the absence
(Fig. 7A) or presence (Fig. 7B) of carbachol, a speciWc
agonist of the muscarinic receptor. Figs. 7A and B show
that carbachol increased calcium Xuorescence of speciWc
cells within the population. We randomly chose 30 cells
that responded to carbachol and determined the average
increase in calcium Xuorescence intensity to be 3.2-fold
(Fig. 7C). The 30 responsive cells were imaged with a
series of emission Wlters, and all were found to contain
the 605-nm QD codes. We also imaged 30 cells each
expressing the serotonin 2A or 2B receptors and found
the average fold-increase in calcium Xuorescence to be
1.0 and 1.1-fold, respectively (Fig. 7C). Control experiments showed that the agonist serotonin had no eVect on
the muscarinic receptor. Thus, these results indicate that
the calcium responses of individual cells can be assayed
within a mixed cell population and that QDs can be used
to encode and multiplex a functional cell-based assay.
Discussion
We have developed a generic method of encoding
cells for use in a variety of assays. The QD codes can be
used with diVerent cell types, and they do not appear to
aVect the cell’s physiology under the assay conditions

that we tested. QD cell codes can be used to potentially

multiplex virtually any microscope- or FACS-based cellular assay with an optical readout. The readout can be
Xuorescence, luminescence, or bright-Weld imaging. For
a Xuorescent reporter, such as an organic dye that senses
intracellular calcium, the codes can be chosen to minimize spectral overlap with the reporter.
Although we demonstrated encoding of calcium,
reporter gene, receptor internalization, and competition
binding assays, QD cell codes could also be used to multiplex other existing assays such as toxicity, apoptosis,
neurite outgrowth, and membrane potential. This encoding platform may also enable new assays where it is
essential that multiple cell types be barcoded and
observed within a complex and mixed population. For
example, certain types of motile tumor cells have been
shown to engulf QDs [33], and it may be possible to use
this process to introduce unique QD codes into multiple
tumor cell types and observe their motility simultaneously within a mixed population.
Another application is based on recent results showing that QDs and DNA can be cotransfected into cells
using cationic lipids (L.C. Mattheakis, unpublished
data). Thus, it may be possible to encode DNA transfections and observe the behavior of the transfected cells in
suspension rather than on an array surface.
Many of these applications are not easily achieved
using organic dyes or other encoding formats. Organic
dyes provide a limited number of color choices, and their
spectral overlap further decreases the number of useful
codes. The photostability of organic dyes varies widely,
and this may complicate their use as cell codes when used
in diVerent combinations [34]. Metallic barcodes, metal
nanoparticles that are self-encoded with submicrometer
stripes, are being developed for a variety of multiplexing
applications, but these particles are submicrometer in

size and much larger than mammalian cells [35].
QDs can potentially create a large number of cell
codes. A simple binary encoding strategy (2N ± 1, N D
number of resolvable colors) estimates 31 distinct codes
using Wve colors. Our results, however, suggest that a
higher encoding capacity is possible because the combination of QD colors 582 and 630 nm yielded distinct
codes when the QD colors were used at diVerent intensity ratios. Therefore, using Wve colors, it may be possible
to generate codes based on ternary (3N ± 1, 243 codes) or
quaternary (4N ± 1, 1023 codes) encoding schemes depending on the optical resolution of the intensity levels.
Although these numbers may seem small compared to
the number of features present on DNA or protein
microarrays, they may be more than adequate for most
cellular assays.
Encoded cells also oVer important advantages
compared to cell arrays. Cell arrays require adherant
cells, and the cells must be grown in parallel and in a
miniaturized format without cross-contamination. The
cells must also have access to the necessary nutrients,


L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

207

Fig. 7. Multiplex calcium assay using QDs. CHO cells expressing the muscarinic M1 receptor (ATCC, Manassas, VA) and serotonin 2A and serotonin 2B receptors (see Materials and methods) were encoded separately with 608-, 582-, and 630-nm QDs, respectively. The cells were mixed, and
aliquots were transferred to the wells of a 96-well clear-bottomed assay plate. The calcium assay and imaging were as described under Materials and
methods. Shown are the same Weld of view images of cells incubated in the absence (A) or presence (B) of 30 M carbachol. Some cells are circled
and identiWed as examples. The images are composites of the calcium dye and QD Xuorescent images. (C) EVect of carbachol on the cellular calcium
response. For each receptor type, the calcium dye Xuorescence intensities of 30 cells within the mixed population was measured before and after addition of carbachol to determine the average fold response. The error bars indicate standard deviation values.

chemical compounds, and macromolecules required for
cell growth under a variety of conditions. Linking the
codes directly to single cells allows the cells to be grown
and analyzed using well-established and routine cell
analysis methods.
In conclusion, we have created a generic method for
encoding single cells. The applications for this technology include multiplexing cell-based assays for drug
screening and studying mixed cell populations. The
ability to tag individual cells with unique Xuorescent
barcodes should greatly facilitate the study of cell/cell
interactions and other complex phenotypes.
Acknowledgments
We thank Stephen Rees (GlaxoSmithKline) for
kindly providing the CHO cell line expressing the
2adrenergic receptor and renilla luciferase reporter gene,
Dr. William Hyun (UCSF Laboratory for Cell Analysis)

for confocal microscopy analysis, and Dr. Mark von
Zastrow (UCSF) for helpful discussions.

References
[1] S.P. Fodor, J.L. Read, M.C. Pirrung, L. Stryer, A.T. Lu, D. Solas,
Light-directed, spatially addressable parallel chemical synthesis,
Science 251 (1991) 767–773.
[2] D.J. Lockhart, H. Dong, M.C. Byrne, M.T. Follettie, M.V. Gallo,
M.S. Chee, M. Mittmann, C. Wang, M. Kobayashi, H. Horton, E.L.
Brown, Expression monitoring by hybridization to high-density
oligonucleotide arrays, Nat. Biotechnol. 14 (1996) 1675–1680.
[3] R.M. de Wildt, C.R. Mundy, B.D. Gorick, I.M. Tomlinson, Antibody arrays for high-throughput screening of antibody-antigen
interactions, Nat. Biotechnol. 18 (2000) 989–994.
[4] A. Lueking, M. Horn, H. EickhoV, K. Bussow, H. Lehrach, G.

Walter, Protein microarrays for gene expression and antibody
screening, Anal. Biochem. 270 (1999) 103–111.
[5] G. MacBeath, S.L. Schreiber, Printing proteins as microarrays for
high-throughput function determination, Science 289 (2000)
1760–1763.


208

L.C. Mattheakis et al. / Analytical Biochemistry 327 (2004) 200–208

[6] K.J. Albert, D.R. Walt, D.S. Gill, T.C. Pearce, Optical multibead
arrays for simple and complex odor discrimination, Anal. Chem.
73 (2001) 2501–2508.
[7] J. Chen, M.A. Iannone, M.S. Li, J.D. Taylor, P. Rivers, A.J. Nelsen,
K.A. Slentz-Kesler, A. Roses, M.P. Weiner, A microspherebased assay for multiplexed single nucleotide polymorphism
analysis using single base chain extension, Genome Res. 10 (2000)
549–557.
[8] T.A. Dickinson, K.L. Michael, J.S. Kauer, D.R. Walt, Convergent,
self-encoded bead sensor arrays in the design of an artiWcial nose,
Anal. Chem. 71 (1999) 2192–2198.
[9] T.M. McHugh, Y.J. Wang, H.O. Chong, L.L. Blackwood, D.P.
Stites, Development of a microsphere-based Xuorescent immunoassay and its comparison to an enzyme immunoassay for the
detection of antibodies to three antigen preparations from Candida albicans, J. Immunol. Methods 116 (1989) 213–219.
[10] K.G. Oliver, J.R. Kettman, R.J. Fulton, Multiplexed analysis of
human cytokines by use of the FlowMetrix system, Clin. Chem. 44
(1998) 2057–2060.
[11] M. Han, X. Gao, J.Z. Su, S. Nie, Quantum-dot-tagged microbeads
for multiplexed optical coding of biomolecules, Nat. Biotechnol.
19 (2001) 631–635.

[12] C.S. Chen, M. Mrksich, S. Huang, G.M. Whitesides, D.E. Ingber,
Micropatterned surfaces for control of cell shape, position, and
function, Biotechnol. Prog. 14 (1998) 356–363.
[13] M. Mrksich, C.S. Chen, Y. Xia, L.E. Dike, D.E. Ingber, G.M.
Whitesides, Controlling cell attachment on contoured surfaces
with self-assembled monolayers of alkanethiolates on gold, Proc.
Natl. Acad. Sci. USA 93 (1996) 10775–10778.
[14] M. Mrksich, L.E. Dike, J. Tien, D.E. Ingber, G.M. Whitesides,
Using microcontact printing to pattern the attachment of mammalian cells to self-assembled monolayers of alkanethiolates on transparent Wlms of gold and silver, Exp. Cell Res. 235 (1997) 305–313.
[15] C.W. Xu, High-density cell microarrays for parallel functional
determinations, Genome Res. 12 (2002) 482–486.
[16] M.N. Yousaf, B.T. Houseman, M. Mrksich, Using electroactive
substrates to pattern the attachment of two diVerent cell populations, Proc. Natl. Acad. Sci. USA 98 (2001) 5992–5996.
[17] J. Ziauddin, D.M. Sabatini, Microarrays of cells expressing
deWned cDNAs, Nature 411 (2001) 107–110.
[18] A.P. Alivisatos, Semiconductor clusters, nanocrystals, and quantum dots, Science 271 (1996) 933–937.
[19] B.O. Dabbousi et al., (CdSe)ZnS Core-shell quantum dots: synthesis and characterization of a size series of highly luminescent
nanocrystallites, J. Phys. Chem. 101 (1997) 9463–9475.
[20] D. Gerion, W.J. Parak, S.C. Williams, D. Zanchet, C.M. Micheel,
A.P. Alivisatos, Sorting Xuorescent nanocrystals with DNA,
J. Am. Chem. Soc. 124 (2002) 7070–7074.

[21] M. Bruchez Jr., M. Moronne, P. Gin, S. Weiss, A.P. Alivisatos,
Semiconductor nanocrystals as Xuorescent biological labels, Science 281 (1998) 2013–2016.
[22] E.R. Goldman, E.D. Balighian, H. Mattoussi, M.K. Kuno, J.M.
Mauro, P.T. Tran, G.P. Anderson, Avidin: a natural bridge for
quantum dot-antibody conjugates, J. Am. Chem. Soc. 124 (2002)
6378–6382.
[23] W.C. Chan, S. Nie, Quantum dot bioconjugates for ultrasensitive
nonisotopic detection, Science 281 (1998) 2016–2018.

[24] S. Pathak, S.K. Choi, N. Arnheim, M.E. Thompson, Hydroxylated
quantum dots as luminescent probes for in situ hybridization,
J. Am. Chem. Soc. 123 (2001) 4103–4104.
[25] S.J. Rosenthal, I. Tomlinson, E.M. Adkins, S. Schroeter, S. Adams,
L. SwaVord, J. McBride, Y. Wang, L.J. DeFelice, R.D. Blakely,
Targeting cell surface receptors with ligand-conjugated nanocrystals, J. Am. Chem. Soc. 124 (2002) 4586–4594.
[26] B. Dubertret, P. Skourides, D.J. Norris, V. Noireaux, A.H. Brivanlou, A. Libchaber, In vivo imaging of quantum dots encapsulated
in phospholipid micelles, Science 298 (2002) 1759–1762.
[27] X. Wu, H. Liu, J. Liu, K.N. Haley, J.A. Treadway, J.P. Larson, N.
Ge, F. Peale, M.P. Bruchez, ImmunoXuorescent labeling of cancer
marker Her2 and other cellular targets with semiconductor quantum dots, Nat. Biotechnol. 21 (2003) 41–46.
[28] J.K. Jaiswal, H. Mattoussi, J.M. Mauro, S.M. Simon, Long-term
multiple color imaging of live cells using quantum dot bioconjugates, Nat. Biotechnol. 21 (2003) 47–51.
[29] M.A. Hines, P. Guyot-Sionnest, Synthesis and characterization of
strongly luminescent ZnS-capped CdSe nanocrystals, J. Phys.
Chem. 100 (1996) 468–471.
[30] M.C. Morris, J. Depollier, J. Mery, F. Heitz, G. Divita, A peptide
carrier for the delivery of biologically active proteins into mammalian cells, Nat. Biotechnol. 19 (2001) 1173–1176.
[31] T. Mosmann, Rapid colorimetric assay for cellular growth and
survival: application to proliferation and cytotoxicity assays,
J. Immunol. Methods 65 (1983) 55–63.
[32] L.J. Stephens, Schaum’s Outline of Beginning Statistics, McGrawHill Trade, New York, 1997.
[33] W.J. Parak, R. Boudreau, M. Le Gros, D. Gerion, D. Zanchet,
M. Micheel, S.C. Williams, A.P. Alivisatos, C. Larabell, Cell
motility and metastatic potential studies based on quantum
dot imaging of phagokinetic tracks, Adv. Mater. 14 (2002) 882–
885.
[34] L.C. Taylor, D.R. Walt, Application of high-density optical microwell arrays in a live-cell biosensing system, Anal. Biochem. 278
(2000) 132–142.
[35] S.R. Nicewarner-Pena, R.G. Freeman, B.D. Reiss, L. He, D.J.

Pena, I.D. Walton, R. Cromer, C.D. Keating, M.J. Natan, Submicrometer metallic barcodes, Science 294 (2001) 137–141.