Unraveling DNA Condensation with Optical Tweezers - Zhuang

P ERSPECTIVES
wafer bonding approach allows point defects
and potentially also waveguides to be
introduced parallel to the layers by proper
lithography. In addition, wafer bonding can
incorporate a layer that acts as a light
source (if optically pumped). In the specific
fabrication approach of Ogawa et al. (3),
the woodpile structure consists of a GaAs
photonic crystal with point defects and an
InGaAsP multi–quantum-well structure as
light emitter. The measured photoluminescence
spectra clearly prove the photonic
band gap. In addition, they show in-gap defect
modes depending on the defect size, as
expected from theory. The authors succeeded
in mapping the defect by the in-gap
frequency of 1.55 µm.
What are the technological implications?
In principle, the combination of lithography,
etching, and wafer bonding
could be extended to more complex photonic
circuits with desirable properties. To
take full advantage of the 3D properties of
photonic crystals, the number of layers has
to be increased. The real question is
whether wafer bonding may offer an economically
and technologically feasible
fabrication approach for complex and
functional 3D photonic crystals containing
waveguides and microresonators. Based
on present knowledge, for economical fabrication
the problem of wafer scale alignment
has to be solved. Another problem is
the cost to etch down all the wafers to
woodpile-layer thickness, a process that
might be too time-consuming for mass
production. The fabrication of siliconbased
structures would be attractive. In
this case, the optically pumped active layer
could be SiO 2 , doped with silicon
nanocrystals and erbium, emitting around
1.5 µm wavelength. The required alignment
of silicon woodpiles appears to be
more difficult than in the case of GaAs,
owing to the lower electronic band gap of
silicon.
Ogawa et al. (3) have described the fabrication
of 3D photonic crystals of woodpile
type including point defects and a multi–quantum-well
structure. They report on
photoluminescence measurements of corresponding
microresonator states within
the band gap. This is a major step forward
in the direction of desirable complex photonic
circuits as integrated “semiconductors
for light.”
References
1. S. John, Phys. Rev. Lett. 58, 2486 (1987).
2. E. Yablonovitch, Phys. Rev. Lett. 58, 2059 (1987).
3. Sh. Ogawa et al., Science 305, 227 (2004); published
online 3 June 2004 (10.1126/science.1097968).
4. K. M. Ho et al., Phys. Rev. Lett. 65, 3152 (1990).
5. A. Blanco et al., Nature 405, 437 (2000).
6. O. Toader, S. John, Science 292, 1133 (2001).
7. Y. V. Miklyaev et al., Appl. Phys. Lett. 82, 1284 (2003).
8. M. Alexe, U. Gösele, Eds., Wafer Bonding—Applications
and Technology (Springer, Berlin, 2004).
MOLECULAR BIOLOGY
Unraveling DNA Condensation
with Optical Tweezers
Since the invention of optical tweezers
by Ashkin, Chu, and co-workers almost
20 years ago (1), this technique
has been applied to a variety of biological
systems. It has yielded a wealth of information,
such as how biopolymers behave
under stress (2), how DNA-interacting enzymes
decode or digest DNA (3–6), how
motor proteins walk along molecular
tracks (7, 8), and how RNA and protein
molecules fold and unfold (9–11). By allowing
the manipulation of individual molecules,
optical tweezers reveal how these
biological systems work at the molecular
level. On page 222 of this issue, Case et
al. (12) demonstrate yet another elegant
application of this technique: unraveling
the mystery of how proteins called condensins
make DNA take on a compact
form.
In essence, optical tweezers use a tightly
focused laser beam to trap a particle in
three dimensions and, through redirection
of the beam, manipulate the particles (1).
Focused light exerts two forces on the particle.
The gradient force draws the particle
toward the focus of the beam where the
The author is in the Department of Chemistry and
Chemical Biology, Department of Physics, Harvard
University, Cambridge, MA 02138, USA. E-mail:
zhuang@chemistry.harvard.edu
Xiaowei Zhuang
light field is the strongest. The scattering
force arises from the radiation pressure exerted
on the particle by absorbed or scattered
photons, which blow the particle
down the optical axis like a wind. When
balanced, these two forces hold the particle
just slightly downstream of the light focus.
Perhaps the most fascinating applications
of optical tweezers are in biology.
Because biological molecules are often too
small to be manipulated directly by optical
tweezers, a micrometer-sized dielectric
sphere is often attached to the molecule of
interest to serve as a handle (see the figure,
panel A). This allows one to manipulate the
position of the attached biomolecule and to
exert a well-defined force. Both parameters
can be determined with great accuracy:
The position of the microsphere can be
measured to within 1 nm, and the force to
within 1 pN. Optical tweezers, therefore,
allow exquisite control over the manipulation
of biomolecules.
Using this technique, Case et al. have
discovered that bacterial condensins compact
DNA in an orderly fashion. DNA condensation
is essential for cell division because
compact DNA is much easier to split
between two daughter cells than DNA in its
expanded form. The chromosomal DNA of
bacteria is compacted into a nucleoid (13),
and eukaryotic cells employ an elaborate
mitotic machinery to achieve chromosome
compaction (14). In both worlds, SMC
(Structural Maintenance of Chromosomes)
proteins are crucial players in the process
of DNA condensation (13, 14).
Case and co-workers concentrated on
the Escherichia coli condensin MukBEF,
which consists of an SMC dimer (MukB)
and two non-SMC subunits (MukE and
MukF) (see the figure, panel A). In their
setup, polystyrene beads were attached to
the two ends of a piece of double-stranded
DNA. One bead was held by an optical trap
and the other by a micropipette (panel A).
By moving the micropipette away from the
optical beam, the investigators obtained a
force-versus-extension curve for the DNA.
In a previous study, this group had used a
similar method to examine the mechanical
properties of DNA. However, when they
applied their trick to DNA in the presence
of MukBEF and ATP, they were caught by
surprise.
First, they found that the force-extension
curve increased more quickly than
observed for naked DNA (see the figure,
panel B). This is perhaps not so surprising
considering that MukBEF condenses
DNA. A very interesting phenomenon,
however, occurred as the force reached 17
pN: A sawtooth pattern became superimposed
on an otherwise flat region of the
curve, indicating that the DNA-MukBEF
complex underwent a phase transition consisting
of individual decondensation
events (see the figure, panel B). As the
force was relaxed, the DNA returned to its
original condensed form. Amazingly,
when the applied force was less than but
still close to 17 pN, recondensation was so
slow that individual condensation steps of
188
9 JULY 2004 VOL 305 SCIENCE www.sciencemag.org

Force (pN)
A
B
40
30
20
10
0
1.0
Dielectric
sphere
Micropipette
Pull 1
Pull 2
Pull 3
1.5 2.0 2.5 3.0
Extension (m)
35 nm were observed, probably caused by
the conformational change of a single
MukBEF molecule. Condensation took
place in the presence of ATP and its nonhydrolyzable
analogs but not in their absence,
indicating that ATP binding rather
than ATP hydrolysis provided the energy
source for condensation.
Perhaps the most surprising result is the
reproducible sawtooth pattern observed
when the same DNA was pulled multiple
times. Sawtooth patterns in force-extension
curves of biopolymers are not a new phenomenon.
When the giant muscle protein,
titin, was stretched by an atomic force apparatus,
a beautiful sawtooth pattern was
observed, indicating the sequential unfolding
of individual immunoglobulin domains
(15). A similar pattern was also seen when
a nucleosome array was pulled by an optical
trap, signaling the release of DNA from
individual histones (16). In these examples,
the sawtooth pattern indicated a trend toward
increasing disruption forces. This
C
MukBEF
DNA
F
F
Optical beam
A nip and tuck for DNA. Unraveling a DNA-MukBEF complex with optical tweezers. (A) In the experimental
setup of Case et al. (12), polystyrene beads (beige spheres) were attached to the two
ends of a piece of double-stranded DNA (purple). One bead was held by an optical trap (orange)
and the other by a micropipette (gray handle). By moving the micropipette away from the optical
beam, the investigators obtained force-versus-extension curves for DNA bound to the bacterial
condensin MukBEF (blue). Here, only the SMC dimers (MukB) are shown; the non-SMC subunits,
MukE and MukF, are believed to bind to the head domains of MukB (blue rectangles). (B) Force-versus-extension
curves obtained for three consecutive pulling runs of the DNA-MukBEF complex. (C)
A model of force-induced decondensation of the DNA-MukBEF complex.
F
P ERSPECTIVES
would be expected given that higher forces
are required to break stronger bonds and so
such bonds tend to break later during the
force loading process. For bonds that require
a similar amount of force to break,
individual disruptions are believed to occur
randomly. However, in the case of the
DNA-MukBEF complex, the investigators
obtained a highly reproducible sawtooth
pattern in consecutive pulling runs with no
correlation between the disruption force of
each peak and its order of occurrence (see
the figure, panel B).
To explain these results, Case and coworkers
propose that ATP-bound MukBEF
polymerizes along the DNA by interactions
between the head domains of adjacent
MukBEF molecules (see the figure, panel
C). This joint-head structure binds tightly
to DNA (panel C) and does not detach
from the DNA even under forces as high as
90 pN. This ATP-dependent interaction between
the two neighboring head domains
of MukBEF and the subsequent stimulation
of binding between DNA and the SMC
dimers is consistent with a recent biochemical
study of the Bacillus subtilis SMC
(17). Case et al. further propose that a
weaker intramolecular interaction between
the two heads of the same MukBEF dimer
causes DNA condensation. The breaking of
this interaction at 17 pN leads to the observed
sawtooth pattern (see the figure,
panel C). The striking reproducibility of
the sawtooth pattern may indicate that this
latter interaction always starts to break
from one end of the condensed DNA filament,
presumably because the force is
weaker for a MukBEF molecule that has
only a single neighbor (panel C).
Although these experiments have provided
new insights into the interactions between
condensins and DNA that are critical
for understanding the DNA condensation
process, much is still unknown. For example,
the compaction ratio observed in this
experiment is only about 4, much less than
the observed value of 10 3 to 10 4 in bacteria.
Concerted actions of condensins, topoisomerases,
other DNA binding proteins, and
DNA supercoiling in vivo are likely to be
responsible for this discrepancy, but the interplay
between these different factors is
largely unknown. Another interesting question
is whether the functional mechanism
of DNA condensation orchestrated by condensins
is conserved between prokaryotic
and eukaryotic cells. Considering the
largely conserved structure of SMC dimers
between the two types of organisms, one
would be inclined to say yes. However, a
recent single-molecule study of Xenopus
laevis condensin I by Strick and co-workers
gave very different results (18). The
most striking difference is that ATP hydrolysis
is required for the compaction of DNA
by condensin I, whereas Case et al. show
that MukBEF-induced DNA condensation
takes place in the presence of nonhydrolyzable
ATP as well. Quantitatively, the step
sizes of individual condensation and decondensation
events are noticeably larger
for condensin I, and so is the overall compaction
ratio observed in the experiment by
Strick and co-workers. Whether these differences
arise from different functional
mechanisms between prokaryotic and eukaryotic
condensins or are merely due to
different experimental conditions is worth
investigating.
Having observed many elegant experiments
on biological molecules using optical
tweezers, I’ve often wondered how far
this method can take us. After all, holding
a molecule at its two ends and then stretching
it is not what happens to biomolecules
naturally in cells. However, the power of
this technique has never ceased to amaze
me when applied with intelligent experi-
www.sciencemag.org SCIENCE VOL 305 9 JULY 2004
189

P ERSPECTIVES
mental design. For example, efforts to improve
the precision of optical traps now allow
the transcription of DNA to be followed
at almost base-pair resolution (19).
Processes as complicated as DNA packaging
by viruses and DNA uptake by bacteria
have been studied under near physiological
conditions (20, 21). Case et al. have now
added another wonderful example to this
list, not to mention the equally impressive
accomplishments attributable to other single-molecule
force spectroscopy techniques
(22–24). In the field of manipulation
and force measurements of single molecules,
it appears that, so far, one is limited
only by one’s imagination.
References
1. A. Ashkin et al., Opt. Lett. 11, 288 (1986).
2. C. Bustamante et al., Science 265, 1599 (1994).
3. M. D. Wang et al., Science 282, 902 (1998).
4. G. J. L. Wuite et al., Nature 404, 103 (2000).
5. R. J. Davenport et al., Science 287, 2497 (2000).
6. T. T. Perkins et al., Science 301, 1914 (2003).
7. A. D. Mehta et al., Nature 400, 590 (1999).
8. K. Svoboda et al., Nature 365, 721 (1993).
9. M. S. Z. Kellermayer et al., Science 276, 1112 (1997).
10. L. Tskhovrebova et al., Nature 387, 308 (1997).
11. B. Onoa et al., Science 299, 1892 (2003).
12. R. B. Case et al., Science 305, 222 (2004); published
online 3 June 2004 (10.1126/science.1098225).
13. D. J. Sherratt, Science 301, 780 (2003).
14. J. R. Swedlow, T. Hirano, Mol. Cell 11, 557 (2003).
15. M. Rief et al., Science 276, 1109 (1997).
16. B. D. Brower-Toland et al., Proc. Natl. Acad. Sci. U.S.A.
99, 1960 (2002).
17. M. Hirano, T. Hirano, EMBO J., 10.1038/sj.emboj.
7600264 (2004).
18. T. R. Strick et al., Curr. Biol. 14, 874 (2004).
19. J. W. Shaevitz et al., Nature 426, 684 (2003).
20. D. E. Smith et al., Nature 413, 748 (2001).
21. B. Maier et al., Nat. Struct. Mol. Biol,10.1038/nsmb783
(2004).
22. T. R. Strick et al., Nature 404, 901 (2000).
23. X. Zhuang, M. Rief, Curr. Opin. Struct. Biol. 13, 88
(2003).
24. A. M. van Oijen et al., Science 301, 1235 (2003).
MATERIALS SCIENCE
Microstructures in 4D
The ability to watch the three-dimensional
growth of a single crystal that
is deeply embedded in a bulk sample,
as Schmidt et al. report on page 229 of this
issue, is a breakthrough in materials characterization
(1). Their work demonstrates
the opportunities created for materials research
by further development of the threedimensional
x-ray diffraction (3DXRD)
microscope at the European Synchrotron
Radiation Facility in Grenoble. The
3DXRD microscope, which Schmidt et al.
developed into a truly 4DXRD microscope
by adding the dimension of time, is likely to
play a major role in revealing the underlying
mechanisms of evolving microstructures
of partially or fully crystalline materials
such as metals, ceramics, and polymers.
Controlling microstructure is important,
because it largely determines the properties
of a wide variety of materials. For example,
the yield strength of metals is inversely proportional
to the average grain size (2), the
conductivity of superconductors is strongly
reduced through grain boundaries (3), and
magnetic domain walls can be pinned by
grain boundaries and precipitates (impurities).
Moreover, the degree of crystallinity
and the size and arrangement of crystallites
in a semicrystalline polymer have a profound
effect on its physical and mechanical
properties (4). Ideally, the formation of the
microstructure should be monitored under
conditions like those of real processing and
in the bulk of the material. This puts high
demands on the measurement technique.
Schmidt et al. have made an exciting step
forward by imaging simultaneously the
The author is in the Department of Materials Science
and Engineering, Delft University of Technology,
Rotterdamseweg 137, 2628 AL Delft, Netherlands. E-
mail: S.E.Offerman@tnw.tudelft.nl
S. Erik Offerman
spatial and time-dependent microstructure
of bulk material down to the micrometer
range, which was not possible until now
with any other experimental technique or
ab initio calculation.
Traditionally, metallurgists use light microscopy
to reveal the microstructure from
cross sections of the material.
However, the 2D
image that is obtained in
this way is an oversimplification
of the complex
3D microstructure. The
combined use of computer-aided
reconstruction
techniques and light
microscopy allows the
creation of 3D images of
limited parts of the microstructure
via repeatedly
grinding off a thin
layer of material and taking
an image (5). Despite
the effort, light microscopy
techniques are
limited to ex situ measurements.
The advent of
electron microscopy techniques
drastically improved
the resolution of
the image to the nanometer
level. In addition, in
situ electron microscopy
measurements have been performed to
capture the evolution of the microstructure
at high temperatures (6). Nevertheless,
free-surface effects that are always present
in 2D imaging techniques during in situ
measurements complicate the analysis
and the ability to draw unambiguous conclusions.
Neutron and synchrotron radiation
facilities, which enable high penetration
even in high electron density materials,
have opened the possibility for nondestructive
imaging of the 3D microstructure.
Yet the time dimension was still lacking
in 3D imaging with micrometer spatial
resolution.
Using the exceptionally high brilliance
of third-generation synchrotron sources,
Schmidt et al. developed a technique to
study microstructures in 4D with a spatial
resolution of micrometers and a time resolution
of minutes. The strength of the
technique that Schmidt et al. developed is
that many relevant aspects of the evolving
Microstructural mapping. Schematic drawing of a deformed metal,
showing the complexity and inhomogeneities of the microstructure:
dislocations (T-shaped symbol), precipitates (small solid
spheres), subgrain and grain boundaries, and a recrystallizing grain in
3D (arrows show direction of growth of grain surfaces). Large color
differences indicate large differences in crystallographic orientation
(top).
microstructure can be measured simultaneously
for the bulk of the material. This
circumvents the need for additional measurements
under different conditions and
of a different part of the material. This is
important because microstructures are
very inhomogeneous, as illustrated in the
figure.
Generally, the microstructure of a polycrystalline
material is composed of multi-
190
9 JULY 2004 VOL 305 SCIENCE www.sciencemag.org