ATP binding and strand-passage.
More speculatively, it would be
fascinating tohave a corresponding
single-molecule method to analyse
ATP hydrolysis simultaneously with
DNA supercoiling, to really dissect
the energy coupling aspects of
these systems.
References
1. Wallace, M.I., Malloy, J.E., and Trentham,
D.R. (2003). Combined single-molecule
force and fluorescence measurements for
biology. J. Biol. 2, 4.
2. Charvin, G., Strick, T.R., Bensimon, D.,
and Croquette, V. (2005). Tracking
topoisomerase activity at the single-
molecule level. Annu. Rev. Biophys.
Biomol. Struct. 34, 201­219.
3. Gore, J., Bryant, Z., Stone, M.D.,
Noš llmann, M., Cozzarelli, N.R., and
Bustamante, C. (2006). Mechanochemical
analysis of DNA gyrase using rotor bead
tracking. Nature 439, 100­104.
4. Champoux, J.J. (2001). DNA topoiso-
merases: structure, function, and mecha-
nism. Annu. Rev. Biochem. 70, 369­413.
5. Bates, A.D., and Maxwell, A. (2005). DNA
Topology(Oxford:OxfordUniversityPress).
6. Bryant, Z., Stone, M.D., Gore, J., Smith,
S.B., Cozzarelli, N.R., and Bustamante, C.
(2003). Structural transitions and elasticity
from torque measurements on DNA.
Nature 424, 338­341.
7. Bates, A.D., O'Dea, M.H., and Gellert, M.
(1996). Energy coupling in Escherichia coli
DNA gyrase: the relationship between
nucleotide binding, strand passage and
DNA supercoiling. Biochemistry 35, 1408­
1416.
8. Reece, R.J., and Maxwell, A. (1991). DNA
gyrase: Structure, mechanism, and
interaction with antibiotics. CRC Crit. Rev.
Biochem. Mol. Biol. 26, 335­375.
9. Ali, J.A., Jackson, A.P., Howells, A.J., and
Maxwell, A. (1993). The 43 kilodalton
N-terminal fragment of the DNA gyrase
B protein hydrolyses ATP and binds
coumarin drugs. Biochemistry 32,
2717­2724.
10. Ali, J.A., Orphanides, G., and Maxwell, A.
(1995). Nucleotide binding to the
43-kilodalton N-terminal fragment of
the DNA gyrase B protein. Biochemistry
34, 9801­9808.
11. Kampranis, S.C., Bates, A.D., and Maxwell,
A. (1999). A model for the mechanism of
strand-passage by DNA gyrase. Proc. Natl.
Acad. Sci. USA 96, 8414­8419.
12. Heddle, J.G., Mitelheiser, S., Maxwell, A.,
and Thomson, N.H. (2004). Nucleotide
binding to DNA gyrase causes loss of the
DNA wrap. J. Mol. Biol. 337, 597­610.
13. Sugino, A., Higgins, N.P., Brown, P.O.,
Peebles, C.L., and Cozzarelli, N.R. (1978).
Energy coupling in DNA gyrase and the
mechanism of action of novobiocin. Proc.
Natl. Acad. Sci. USA 75, 4838­4842.
14. Maxwell, A., and Gellert, M. (1984). The DNA
dependence of the ATPase activity of DNA
gyrase. J. Biol. Chem. 259, 14472­14480.
15. Tingey, A., and Maxwell, A. (1996). Probing
the role of the ATP-operated clamp in the
strand-passage reaction of DNA gyrase.
Nucl. Acids Res. 24, 4868­4873.
School of Biological Sciences, University
of Liverpool, Biosciences Building,
Crown Street, Liverpool L69 7ZB, UK.
E-mail: bates@liv.ac.uk
DOI: 10.1016/j.cub.2006.02.029
+ 2 ATP
­ 2 ADP
­ 2 Pi
­ 2 ADP
­ 2 Pi
­ 2 ADP
­ 2 Pi
* *
* *
Wrapping
Strand-
passage
rotor
= 720°
1
2
3
4
5
7
6
+
* ** *
Wrapping
Wrapping
RL RL
Current Biology
Figure 3. DNA gyrase mechanistic scheme.
A basic reaction cycle for gyrase under normal conditions is shown on the right (2­5).
Only the half wrap including the T segment is shown for clarity. Binding of ATP cap-
tures a T segment (red) (2/3), which may proceed to strand passage (3/4), or be
freed by ATP hydrolysis and product release (3/2). After strand-passage and T seg-
ment exit from the complex (4/5), the clamp is opened on hydrolysis and rate-limiting
(RL) product release (5/2). This cycle introduces two negative supercoils, and results
in a rotor bead rotation of 720 in the single molecule experiment. Increasing tensile
force will affect the wrapping/unwrapping equilibria, stabilising the unwrapped com-
plexes (6 and 7), and reducing the processivity of the reaction by DNA dissociation
from 7 or 2. Product release is still the rate-limiting step, independent of force. Tensile
force will perturb the equilibria as indicated by the red arrows, resulting in an increase
in the efficiency of the strand-passage reaction (3­4 equilibrium).
Cognitive Neuroscience:
Trickle-Down Theories of Vision
The visual cortex is not a passive recipient of information: predictions
about incoming stimuli are made based on experience, partial
information and the consequences of inferences. A combination of
imaging studies in the human brain has now led to the proposal that the
orbitofrontal cortex is a key source of top-down predictions leading to
object recognition.
Jacinta O'Shea1
and
Vincent Walsh2
The study of top-down processing
often reveals a paradoxical feature
of how we theorise about vision.
It seems that in order to explain
the workings of the brain we
sometimes adopt the logic of
Intelligent Design -- smart things
can only be made by, well, by
smarter things. Every visual
neuroscientist has had the
experience, at the end of a lecture,
of being asked by an engineer in
the audience ``why would anyone
design a system like that?'' (you
know it's an engineer because they
Current Biology Vol 16 No 6
R206
always preface the question with
``I am an engineer''). The answer, of
course, is that nobody designed
the system -- it evolved. Yet when
thinking about top-down control,
the dominant assumption is that
there must be a high-level `smart'
area that guides the less smart
sensory cortex.
This is the thinking behind the
idea that the parietal or prefrontal
cortices are important in top-down
visual control -- they are higher
level areas that can somehow
direct visual cortical processes
[1,2]. The same logic guided
initial theorising about cognitive
contours (Figure 1). It was argued
that these contours are perceived
because the brain makes
inferences about what kinds of
objects are consistent with the
retinal images with which it is
presented. Such an inferential
process could only be carried
out by a smart area, and it was
therefore assumed that the
inferotemporal cortex, in some
senses the pinnacle of the visual
system, must be the source of
the perception of cognitive
contours -- and neurons
responding to them were duly
found. Subsequent studies,
however, were able to show that
the architecture of visual areas V2
[3] and V1 [4] could be sufficient
to generate these illusory percepts:
could the visual cortex be smart
after all?
A new study [5] of visual
recognition processes lends
support to the idea that predictive
visual processing follows a trickle-
down path from frontal cortex to
lower level visual areas. Previously,
Bar et al. [6] had observed that
object recognition elicited activity
not only in the temporal cortex, as
one might expect, but also in the
orbitofrontal cortex, an area not
traditionally associated with object
recognition processes. In the new
study, Bar et al. [5] presented
subjects with line drawings or
photographs of animals and
everyday objects and subjects
pressed a key if they recognised
the object. Subjects underwent two
types of brain imaging: functional
magnetic resonance imaging
(fMRI) to identify the areas
activated by high or low spatial
frequencies in the recognition task;
and magnetoencephalography
(MEG) to determine when these
areas were activated.
The study has three core
findings: First, the MEG data
showed that the right orbitofrontal
cortex was activated 50
milliseconds before the right
fusiform gyrus and 85 ms before
the left fusiform gyrus (Figure 2).
Second, activity in the orbitofrontal
cortex and activity in visual areas
in the temporal (fusiform gyrus)
and occipital (visual cortex) lobes
was time locked on a trial-by-trial
basis. And third, orbitofrontal
cortex activity was found to differ
in response to high versus low
spatial frequency stimuli (Figure 3).
On the basis of these data, Bar
et al. [5] propose that the
orbitofrontal cortex ``sensitises the
representation of the most likely
candidate objects in the temporal
cortex as a predictive initial guess''.
The idea is that low spatial
frequency information reaches the
orbitofrontal cortex before it
reaches the visual recognition
areas, and that the orbitofrontal
cortex sends a guiding signal to the
visual recognition areas that says
something like ``here's a gistimate
(sic) of what you're looking for''.
Our engineering friend already
has both hands in the air to ask the
usual question, but perhaps we can
also see a biologist wanting to ask
a different question: why would
something evolve like this?
Trickle-down theories are based on
movement of resources from the
rich to the poor and in the case of
the orbitofrontal cortex, when it
comes to cognitive functions we
seem to be talking about the
super-rich. As Bar et al. [5] note,
the orbitofrontal cortex is not
traditionally associated with visual
recognition processes, but it is
associated with emotion, reward-
association, impulsivity and
decision making [7,8]. It is also an
area that receives inputs from all
the senses. What we might
consider, then, is whether the
orbitofrontal cortex performs some
function common to all its sensory
inputs or whether its role in
top-down visual control may be
a special case.
If one were to look for an
explanation of the orbitofrontal
cortex activations in terms of what
is already known about this area,
some alternative explanations
begin to present themselves. The
orbitofrontal cortex is
indispensable for forming normal
stimulus­reward associations [8].
Hence it is possible that the
orbitofrontal cortex activations
seen in this study represent
a prediction about the likelihood
that one will identify an object
correctly. In some conditions,
subjects were required to indicate
their level of knowledge about the
object. Thus, the activations may
be representative of subjects'
emerging confidence in their
performance on any given trial,
and they could evolve in parallel
and independently of visual
predictive processes. Such
a confidence judgement might also
Current Biology
Figure 1. Two examples of
`cognitive' contours.
The lines forming the trian-
gle and the long lines divid-
ing the three sets of shorter
horizontal lines are both il-
lusory.Initial explanations of
these percepts suggested
that they must have been
constructed by higher visual
areas such as the inferotem-
poral cortex but we now
know that V1 and V2 re-
sponses may produce these
illusions.
Dispatch
R207
correlate with recognised versus
unrecognised objects on a trial by
trial basis. Another recent study [9],
for example, has shown that activity
in orbitofrontal cortex can correlate
with a subject's level of uncertainty
when making a decision.
Another key question raised by
the new work of Bar et al. [5] is
whether activations in other areas
that preceded orbitofrontal cortex
activation could also reflect top
down processing. There was
greater activation for recognised
versus non-recognised objects
both in the early visual cortex
(Figure 2) and in the right frontal
eye fields (Figure 2). There is
compelling evidence that the
frontal eye fields play a role in
top-down visual processing. The
frontal eye fields respond as early
as 40 milliseconds after a stimulus
is presented [10], have generated
a representation of target location
by 100­130 milliseconds [11], and
share topographically organised
connections with a wide range of
visual areas [12]. They are also
closer than the orbitofrontal cortex
to the visual areas and therefore
satisfy the constraint -- or at least
the common observation -- that
evolution, like an engineer, likes
to keep its wires short to limit the
possibility of error [13].
There is a further reason to
emphasise the role of interactions
between the frontal eye field and
the visual cortex. The visual areas
connected to the frontal eye fields
encode both low and high spatial
frequency information, are closely
connected to the fusiform
recognition areas, and also
respond to visual stimuli before
the orbitofrontal cortex. It seems
entirely possible that the activity
in these areas preceding the
orbitofrontal cortex activity
(Figure 2) encapsulates all the
information required for purely
visual purposes prior to the
fusiform activation, and that the
source of the orbitofrontal cortex
activity may be a parallel process
representing confidence in the
accuracy of the perceptual
judgement that is being made
under ambiguous conditions.
The work of Bar et al. [5] shows
that even simple tasks can yield
surprising and intriguing results
and that understanding the
temporal relationships between
distant regions of cortex is a key
to a more sophisticated
understanding of the pathways
from sense to action. The dual
technique assault, adopted by
Bar et al. [5], combining fMRI
and MEG, provides a richness
of data that will invite new
interpretations -- which we hope
will be a flood rather than a trickle.
References
1. Hung, J., Driver, J., and Walsh, V. (2005).
Visual Selection and Posterior Parietal
Cortex: Effects of repetitive transcranial
magnetic stimulation on partial report
analyzed by Bundesen's theory of
visual attention. J. Neurosci. 19,
9602­9612.
2. Peers, P.V., Ludwig, C.J.H., Rorden, C.,
Cusack, R., Bonfiglioli, C., Bundesen, C.,
Driver, J., Antoun, N., and Duncan, J.
(2005). Attentional functions of the parietal
and frontal cortices. Cerebr. Cortex 15,
1469­1484.
3. Peterhans, E., and von der Heydt, R.
(1989). Mechanisms of contour
perception in monkey visual cortex. II.
Contours bridging gaps. J. Neurosci. 9,
1749­1763.
4. Grosof, D.H., Shapley, R.M., and Hawken,
M.J. (1993). Macaque V1 neurons can
signal `illusory' contours. Nature 365,
550­552.
5. Bar, M., Kassa, K.S., Ghuman, A.S.,
Boshyan, J., Schmidt, A.M., Dale, A.M.,
Hamalainen, M.S., Marinkovic, K.,
Schacter, D.L., Rosen, B.R., et al. (2005).
Top-down facilitation of visual recognition.
Proc. Natl. Acad. Sci. USA 103,
449­454.
Figure 2. Trickle up or
trickle down?
Time course of activity in
the experiment reported re-
cently by Bar et al. [5]. The
first areas to be activated
are the right frontal eye
fields (FEF) and early visual
cortex (VC). The second
phase of activity involves
the orbitofrontal cortex
(OFC) and later phases in-
volve the fusiform gyri (FG)
and the right orbitofrontal
cortex. Trickle-down: Bar
et al. [5] suggest that the
orbitofrontal cortex activity
(130 milliseconds) provides
an initial representation
based on low spatial fre-
quency information, which
then guides object identifi-
cation in the fusiform gyri.
Trickle-up: Alternatively, the frontal eye field and visual cortex activity (100 milliseconds)
may extract gist information and convey this to the fusiform gyri while, in parallel, the or-
bitofrontal cortex may be using low spatial frequency information to estimate the likeli-
hood of making a correct perceptual judgment. (Modified from supplementary informa-
tion in [5].)
Figure 3. High and low spatial frequency information.
Images contain different spatial frequencies. Low spatial frequencies carry `broad
brush' information (centre image) and high spatial frequencies carry outline and fine
contour information (far right image). The initial gist of a scene may be dominated by
low frequency information. From the new Bar et al. [5] study the proposal is that low
spatial frequency information reaches the orbitofrontal cortex via fast feedforward con-
nections and that the information is then used to generate predictions about the more
detailed higher spatial frequency information coming in through the visual cortex.
(Figure courtesy of Elliot Freeman.)
Current Biology Vol 16 No 6
R208