Programmable cell adhesion encoded by DNA hybridization

Article abstract of DOI:10.1002/anie.200502421

(Figure Presented) Stuck-on genes: Live cells functionalized with single-stranded DNA oligonucleotides bind, independent of native adhesion mechanisms, to substrates that bear complementary DNA strands in a sequence-specific manner. This strategy has been

Full text of DOI:10.1002/anie.200502421

Angewandte
Chemie
DOI: 10.1002/anie.200502421
Communications
Live cells functionalized with sin-
gle-stranded DNA oligonucleotides bind
to substrates that bear complementary DNA strands in a
sequence-specific manner. This binding is independent of the native
adhesion mechanisms. The strategy has been successfully utilized with
several common mammalian cell lines and device platforms. For more
information see the Communication by M. B. Francis and co-workers
on the following pages.
896
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 896 – 901

Angewandte
Chemie
DNA Recognition
surfaces have typically been patterned with integrin-binding
ligands, such as laminin or fibronectin, or synthetic RGD-
containing peptides (RGD = arginine, glycine, and aspartic
acid) that are derived from them.[1,6] A particular advantage
of this approach is its generality as most adherent cell types
use integrins as their primary adhesion receptors.[7] However,
the commonality of this mechanism makes it difficult to form
patterns in which multiple types of cells are arrayed with
precision on a single surface. As a result, considerable
attention has been directed toward the development of new
technologies for the programmable attachment of cells to
material surfaces.[8,9]
Herein we show that cell adhesion events can be
programmed through the attachment of synthetic single-
stranded DNA (ssDNA) strands to the surfaces of living cells.
These strands are then used to anchor cells to specified
locations on surfaces in a sequence-dependent fashion. This
provides an attractive means by which to control the adhesion
properties of living cells without a dependence on the
DOI: 10.1002/anie.200502421
Programmable Cell Adhesion Encoded by DNA
Hybridization**
Ravi A. Chandra, Erik S. Douglas, Richard A. Mathies,
Carolyn R. Bertozzi, and Matthew B. Francis*
Microfabrication and patterning techniques from the semi-
conductor industry have yielded promising new tools for the
study of cell biology,[1] foreshadowing future opportunities for
cell-based devices, including biosensors,[2] drug-screening
platforms,[3] artificial tissues,[4] and designed networks of
neurons.[5] To date, the attachment of cells to surfaces has
been achieved by exploiting the cellꢀs natural repertoire of
adhesion receptors, with a particular emphasis on the
integrins.[6] To achieve cell adhesion in these systems, cognate
[*] Prof. M. B. Francis
Department of Chemistry
University of California, Berkeley
and
Materials Sciences Division
Lawrence Berkeley National Laboratory
Berkeley, CA 94720 (USA)
Fax: (+1)510-643-3079
E-mail: francis@cchem.berkeley.edu
R. A. Chandra
Department of Chemistry
University of California, Berkeley
Berkeley, CA 94720 (USA)
E. S. Douglas
UCB/UCSF Joint Graduate Group in Bioengineering
University of California, Berkeley
Berkeley, CA 94720 (USA)
Prof. R. A. Mathies
Department of Chemistry and
UCB/UCSF Joint Graduate Group in Bioengineering
University of California, Berkeley
and
Physical Biosciences Division
Lawrence Berkeley National Laboratory
Berkeley, CA 94720 (USA)
Prof. C. R. Bertozzi
Departments of Chemistry & Molecular andCell Biology
and
HowardHughes Medical Institute
University of California, Berkeley
and
Materials Sciences Division
Lawrence Berkeley National Laboratory
Berkeley, CA 94720 (USA)
[**] The authors thank H. Lee, J. Prescher, M. Hangauer, andN. Toriello
for helpful discussions and assistance. R.A.C. and E.S.D. were
supported by National Science Foundation Predoctoral Fellowships.
This work was supportedby generous financial support from the
Nanoscale Science, Energy, andTechnology (NSET) Program of the
Department of Energy.
Figure 1. Covalent attachment of ssDNA to cell surfaces. a) Azides can be
delivered to cell-surface glycans thorough metabolism of Ac4ManNAz to
SiaNAz. b) The Staudinger ligation of a functionalized phosphine and an azide
results in the formation of an amide bond. c) Phosphine-bearing ssDNA strands
are synthesizedfrom amino ssDNA andphosphine–PFPs. Ac 4ManNAz=acetyl-
ated-N-a-azidoacetylmannosamine, SiaNAz=N-a-azidoacetyl sialic acid,
PFP=pentafluorophenyl.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2006, 45, 896 –901
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
897

Communications
receptors that they possess. An advantage of DNA-based
duce the azide groups. The cells were then reacted with
phosphine–DNA conjugates that were prehybridized with
complementary strands that bore a biotin reporter tag for a
period of 1–2 h (Figure 2a). After the unreacted phosphine–
DNA was removed through rinsing, the cells were treated
with fluorescein isothiocyanate-labeled avidin (avidin-FITC).
Flow cytometry analysis established the presence of DNA
bound to the cell surface (Figure 2b). From these data, we
estimate the number of surface DNA conjugates to be
approximately 270000 per cell under the conditions used.
This value represents a lower limit as it is unlikely that each
step of the detection assay (prehybridization, cell-surface
conjugation, and avidin-FITC labeling) proceeded with a
quantitative yield.[15] Similar results were obtained for Jurkat
(human T-cell line) cells. Control experiments, carried out by
using DNA strands that lackreactive phosphines, showed no
detectable binding above background levels. Similarly,
experiments conducted using azide-free cells, but with
systems is the virtually unlimited number of possible coding
specificities as was demonstrated in previous studies that have
directed synthetic lipid vesicles to supported lipid bilayers
with complementary sequences.[10] It is envisioned that, when
combined with DNA-patterning techniques developed for
microarray platforms,[11] this strategy will allow increasingly
complex networks of living cells to be created through self-
assembly.
Although there are well-established methods for the
functionalization of devices with ssDNA,[11,12] the attachment
of ssDNA to the surface of living cells has not yet been
reported. To do this,[13] we have developed a synthetic method
based on the introduction of specific chemical handles onto
cell surfaces by using metabolic oligosaccharide engineering.
Briefly, this was achieved though the incorporation of azides
into cell surface glycoconjugates by the administration of
peracetylated N-a-azidoacetylmannosamine (Ac4ManNAz).
Previous studies have shown
that this unnatural sugar is
metabolized to the correspond-
ing N-a-azidoacetyl sialic acid
(SiaNAz) within membrane-
O
a)
O
P
OCH3
H
O
O
HO
N
CO2
OH
HO
O
PPh2
5
associated
glycans
(Fig-
O
O
O
H
ure 1a).[14]
N
N3
HO
Phosphine-reporter duplex
We have reported previ-
ously the use of the Staudinger
ligation (Figure 1b) as a bio-
orthogonal reaction for the
modification of cell-surface
azides with phosphine reagents,
yielding stable amide link-
ages.[15] For these studies, we
have expanded this technique to
include oligonucleotide attach-
ment.[16] We prepared modified
ssDNA with a phosphine group
through the reaction of a 5’-
amine-modified ssDNA with a
phosphine pentafluorophenyl
(PFP) ester, as shown in Fig-
ure 1c. Model studies con-
ducted with simple azides con-
firmed that the phosphine–
ssDNA conjugate undergoes
the Staudinger ligation to form
the expected amide-linked
product (see Supporting Infor-
mation). Little or no oxidation
of the phosphine–DNA conju-
gates was observed during han-
dling or storage.
O
HO
CO2
OH
HO
O
O
O
H
N
O
N
HO
H
H
O
O
O
P
N
O
PPh2
5
O
O
b)
Control
Control
160
DNNAA
Phhoossphpihneine
120
80
DNA
DNA
40
0
To verify the ability of the
100
101
102
103
phosphine–DNA conjugates to
undergo ligation to cell surfaces,
Fluorescence Intensity
Figure 2. dsDNA complexes are covalently attached to the surface of live cells. a) Experimental overview
for the quantification of exogenous DNA on the cell surface. Complementary biotinylatedreporter
strands are hybridized to the phosphine-bearing DNA. Biotinylated DNA strands are detected after
binding to avidin-FITC. b) Flow cytometry analysis of human embryonic kidney (HEK) cells treated with
biotinylatedDNA strands that lack (blue) or bear (orange) reactive phosphine groups.
human
embryonic
kidney
(HEK) cells were first treated
with Ac4ManNAz in culture
media for three days to intro-
898
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 896 –901

Angewandte
Chemie
phosphine–DNA, showed detectable but greatly reduced
labeling efficiency.
standard photolithographic techniques (see Supporting Infor-
mation). After treatment of the pads with commercially
obtained 5’-thiolated ssDNA strands, a prefabricated channel
layer of (poly)-dimethylsiloxane (PDMS) was affixed on top
to form enclosed chambers. The resultant channels were 6 cm
long, 600 mm wide, and 32 mm deep, which corresponds to a
total volume of 1.5 mL. An external syringe pump was
attached to control solution delivery at rates between 1–
100 mLminÀ1, and a mounted epifluorescence microscope
with a digital camera was used for analysis. Two complemen-
tary 20-mer DNA-sequence pairs (A/A’ and B/B’) were
chosen such that they possessed an identical overall base
composition, but no appreciable affinity for each other.
Strands A’ and B’ were bound to the Au surfaces within the
channels, and strands A and B were attached to the cells by
using the Staudinger ligation method described above. The
presence and hybridization ability of the surface-immobilized
DNA strands in the devices were determined through
incubation with fluorescently labeled probe strands (Fig-
ure 3b).
In parallel, experiments were carried out by using
phosphine–ssDNA conjugates. Although these species are
likely to react with surface azides with rates similar to those
observed for double stranded DNA (dsDNA), attempts to
quantify these groups through subsequent hybridization to
biotinylated DNA were unsuccessful. This is likely due to the
slow hybridization kinetics associated with the low concen-
trations of DNA on the plasma membrane, resulting in
incomplete labeling. As noted by others,[17] the kinetics of
DNA hybridization are substantially different at material
surfaces owing to the dramatically higher local DNA concen-
tration. Thus, we chose instead to confirm the presence of
ssDNA on the cell membranes through surface attachment.
As an experimental platform, microfluidic devices were
constructed to create controlled environments for the intro-
duction of cells and rinsing solutions (Figure 3a).[18] First, a
series of Au pads was patterned on glass surfaces by using
To test the system, we introduced azides onto the
surface of Jurkat cells, a naturally non-adherent cell
type. The labeled cells were then divided into two
subsets and each of these was modified with either DNA
sequence A or B. For tracking purposes, the two subsets
of cells were also labeled with either blue (7-amino-4-
chloromethylcoumarin) or green (5-chloromethylfluor-
escein diacetate) cytosolic dyes. Equal numbers of each
labeled subpopulation were then mixed and introduced
into the device through the syringe pump. The cells were
incubated for 35 min under static conditions to allow
hybridization, and then unbound cells were flushed from
the system with excess phosphate-buffered saline (PBS)
solution.
Representative fluorescence microscopy data for
these experiments are shown in Figure 4. Only cells
bearing ssDNA strands that were complementary to the
immobilized DNA were observed to bind to the surface
(Figure 4b), whereas otherwise identical cells bearing
mismatched sequences were washed away. Quantitative
binding data obtained from three device regions are
shown in Figure 4c. We found that 95% and 91% of the
bound cells had recognized the appropriate complemen-
tary sequence for the A’ and B’ pads, respectively.
Control experiments with cells that lacksurface DNA
showed no binding.
As further confirmation of the DNA dependence of
our immobilization method, DNA-coated pads were
blocked with complementary DNA sequences prior to
the introduction of labeled cells. Although many cells
bearing complementary DNA strands were exposed to
the pads, virtually no binding was observed following the
standard rinsing procedure (Figure 4d).
A particular advantage of this method is the ability
to immobilize naturally non-adherent cells as well as
adherent ones. This is because metabolic engineering
can be used to introduce azide functional groups on
virtually any cell type that displays sialic acid residues on
its surface.[14] We have explored the generality of this
Figure 3. Construction of a microfluidic device for DNA display. a) Thiolated
ssDNA strands are bound to patterned Au pads located in microfluidic
chambers, solution delivery and washing can be controlled. Coating DNA
strands were A’: 5’-AGT GAC AGC TGG ATC GTT AC-3’ and B’: 5’-CCC TAG
AGT GAG TCG TAT GA-3’. The presence of DNA was detected by using
matched( A, red) and mismatched (B, green) probe strands. b) Fluorescence
images of 1) Probe A incubatedwith A’-coatedsurface; 2) Probe B incubated
with A’-coatedsurface; 3) Probe A incubatedwith B’-coatedsurface.
Angew. Chem. Int. Ed. 2006, 45, 896 –901
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
899

Communications
Figure 5. A single human embryonic kidney (HEK) cell immobilized on
a 16-mm by 16-mm square Au pad. Cells bearing surface DNA strands
were bound to Au pads decorated with complementary strands. Bright
fieldmicroscopy was usedto visualize the cells (magnifiedportion,
40”). Adhesion of cells lacking DNA or bearing noncomplementary
DNA was not observed. Immobilized cells appeared morphologically
similar to unboundcells.
method,[19] we calculated that bound cells are stable to shear
forces that exceed 31 dyncmÀ2, which is far greater than that
needed for most applications.[20] Similar results were noted for
Jurkat and CHO cells.
For this method to be broadly useful, the longevity of cell
surface–DNA complexes must extend beyond the time scale
of the experiment.[1] Although it might be expected that there
would be a loss of surface DNA owing to cell membrane
recycling or secreted nuclease activity,[21] the cells have
remained bound to the surfaces for periods exceeding 25 h
despite the fact that we have taken no measures to avoid these
degradation pathways.
Finally, we evaluated the prolonged loyalty of bound cells
to the surface and their long-term viability though two
additional studies conducted in parallel. In the first, Jurkat
cells labeled with sequence A were immobilized as previously
described, and a gentle flow of Jurkat culture medium was
continuously applied to the device for 25 h. Cells were
visualized by using light microscopy at several time points
to determine their positions. Virtually all of the bound cells
remained at their initial locations for the entire duration of
the experiment. Secondly, we investigated cell viability and
membrane integrity by using redoxsensor red CC-1, a
fluorescent dye that is retained and activated in live cells.
To more closely approximate typical culture conditions, the
PDMS channel layer was removed from the device prior to
introduction of the cells. After the cells had been allowed to
bind for 35 min, Jurkat culture medium was added and the
device was incubated at 378C for 25 h. The cells in the device
were then treated with redoxsensor red CC-1 for 30 min, and
the device was gently rinsed to identify the unbound cells.
Analysis of both the bound and unbound cells indicated that
79% of the DNA-bound cells remained viable after the 25 h
period. This percentage was identical to that observed for the
unbound Jurkat cells. We therefore believe that this method
will be useful for applications that require extended cell
immobilization.[1] More detailed studies of the physiological
effects of DNA-based cell immobilization are currently
underway.
Figure 4. Immobilization of cells is DNA sequence-specific. a) Blue
andgreen dye-labeledJurkat cell populations were functionalizedwith
DNA sequence A or B, respectively. An equal number of cells were
combinedandthe mixture was introducedinto the microfluidic device
(as in Figure 3). b) Fluorescence images of: 1) the cell mixture
prewash; 2) an A’-coatedpadpostwash; 3) a B’-coatedpadpostwash.
c) Summary data for the fraction of the bound cells with either
sequence A (blue), or sequence B (green) under each pad condition.
d) B’-coatedpads were pre-exposedto excess soluble strand B; these
data summarize the fraction of green (sequence B bound) cells bound
before and after washing. Error bars represent the standard deviation
of three separate areas of the device.
technique through the immobilization of several additional
cell lines, including Chinese Hamster Ovary (CHO) and HEK
cells. Both CHO and HEK cells are naturally adherent cell
types and each was successfully immobilized through this
approach. Furthermore, we have found that this attachment
method can be utilized on micropatterned Au pads that
possess the same spatial dimensions as the cells (Figure 5).
The strength of DNA-mediated binding was determined
by subjecting immobilized cells to increasing lateral shear
force. PBS solution was introduced at increasing flow rates
(1–50 mLminÀ1 over a period of 10 min) and the bound HEK
cells were observed. In almost every instance, the bound cells
were not washed away. Through the use of the Navier-Stokes
In summary, we have described a new approach for
engineering cell-adhesion events. This DNA-based strategy is
sequence specific and applicable to a variety of devices and
900
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 896 –901

Angewandte
Chemie
[20] a) M. Takagi, K. Shiwaku, T. Inoue, T. Shirakawa, Y. Sawa, H.
cell types. Current workis focused on tuning the adhesive
properties and studying the long-term behavior of DNA-
immobilized cells, and ongoing efforts are capitalizing on the
programmability of this method to form complex, multi-
cellular patterns for sensing applications.
Matsuda, T. Yoshida, J. Artif. Organs 2003, 6, 222; b) B. T.
Marshall, M. Long, J. W. Piper, T. Yago, R. P. McEver, C. Zhu,
Nature 2003, 423, 190.
[21] a) D. A. Nauman, C. R. Bertozzi, Biochim. Biophys. Acta 2001,
1568, 147; b) K.-C. Hsia, C.-L. Li, H. S. Yuan, Curr. Opin. Struct.
Biol. 2005, 15, 126.
Received: July 12, 2005
Revised: November 7, 2005
Published online: December 21, 2005
Keywords: cell adhesion · chemical biology · DNA recognition ·
.
glycoproteins · Staudinger ligation
[1] a) R. Singhvi, A. Kumar, G. P. Lopez, G. N. Stephanopoulos,
D. I. C. Wang, G. M. Whitesides, D. E. Ingber, Science 1994, 264,
696; b) C. S. Chen, M. Mrksich, S. Huang, G. M. Whitesides,
D. E. Ingber, Science 1997, 276, 1425; c) C. Roberts, C. S. Chen,
M. Mrksich, V. Martichonok, D. E. Ingber, G. M. Whitesides, J.
Am. Chem. Soc. 1998, 120, 6548; d) A. M. P. Turner, N. Dowell,
S. W. P. Turner, L. Kam, M. Isaacson, J. N. Turner, H. G.
Craighead, W. Shain, J. Biomed. Mater. Res. 2000, 51, 430;
e) N. M. Toriello, E. S. Douglas, R. A. Mathies, Anal. Chem
2005, 77, 6935.
[2] a) J. B. Shear, H. A. Fishman, N. L. Allbritton, D. Garigan, R. N.
Zare, R. H. Scheller, Science 1995, 267, 74; b) T. H. Rider, M. S.
Petrovick, F. E. Nargi, J. D. Harper, E. D. Schwoebel, R. H.
Mathews, D. J. Blanchard, L. T. Bortolin, A. M. Young, J. Chen,
M. A. Hollis, Science 2003, 301, 213.
[3] a) J. Ziauddin, D. M. Sabatini, Nature 2001, 411, 107; b) O. E.
Beske, S. Goldbard, Drug Discovery Today 2002, 7, S131.
[4] a) A. Revzin, P. Rajagopalan, A. W. Tilles, F. Berthiaume, M. L.
Yarmush, M. Toner, Langmuir 2004, 20, 2999; b) E. Cukierman,
R. Pankov, D. R. Stevens, K. M. Yamada, Science 2001, 294,
1708.
[5] D. Kleinfeld, K. H. Kahler, P. E. Hockberger, J. Neurosci. 1988,
8, 4098.
[6] a) C. S. Chen, M. Mrksich, S. Huang, G. M. Whitesides, D. E.
Ingber, Biotechnol. Prog. 1998, 14, 356; b) M. Mrksich, Curr.
Opin. Chem. Biol. 2002, 6, 794.
[7] a) R. O. Hynes, Cell 1992, 69, 11; b) F. G. Giancotti, E. Ruo-
slahti, Science 1999, 285, 1028.
[8] M. Kato, M. Mrksich, J. Am. Chem. Soc. 2004, 126, 6504.
[9] C. C. Co, Y.-C. Wang, C.-C. Ho, J. Am. Chem. Soc. 2005, 127,
1598.
[10] a) C. Yoshina-Ishii, G. P. Miller, M. L. Kraft, E. T. Kool, S. G.
Boxer, J. Am. Chem. Soc. 2005, 127, 1356; b) C. Yoshina-Ishii,
S. G. Boxer, J. Am. Chem. Soc. 2003, 125, 3696.
[11] A. C. Pease, D. Solas, E. J. Sullivan, M. T. Cronin, C. P. Holmes,
S. P. A. Fodor, Proc. Natl. Acad. Sci. USA 1994, 91, 5022.
[12] L. M. Demers, D. S. Ginger, S.-J. Park, Z. Li, S.-W. Chung, C. A.
Mirkin, Science 2002, 296, 1836.
[13] All materials and methods may be found in the Supporting
Information.
[14] D. H. Dube, C. R. Bertozzi, Curr. Opin. Chem. Biol. 2003, 7, 616.
[15] E. Saxon, C. R. Bertozzi, Science 2000, 287, 2007.
[16] For a related study that utilizes oligonucleotides and the
Staudinger ligation, see: C. C.-Y. Wang, T. S. Seo, Z. Li, H.
Ruparel, J. Ju, Bioconjugate Chem. 2003, 14, 697.
[17] For an example, see: M. F. Hagan, A. K. Chakraborty, J. Chem.
Phys. 2004, 120, 4958.
[18] D. C. Duffy, J. C. McDonald, O. J. A. Schueller, G. M. White-
sides, Anal. Chem. 1998, 70, 4974.
[19] R. B. Bird, W. E. Stewart, E. N. Lightfoot, Transport Phenom-
ena, Wiley, New York, 1966, pp. 80 – 81.
Angew. Chem. Int. Ed. 2006, 45, 896 –901
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
901