1) Researchers used CT scans and finite element modeling to create a highly accurate 3D biomechanical model of the Allosaurus skull. 2) The model showed that forces during feeding were distributed throughout the skull. Bite forces were surprisingly low compared to estimates for Tyrannosaurus rex, indicating Allosaurus had a different feeding strategy of inflicting rapid bites to soft tissues. 3) Contrary to views of its skull being delicate, the analysis found the Allosaurus skull was strongly built to absorb forces from prey impacts and tooth dragging, though its exact adaptations require more study.

1. news and views
forces and stresses were transmitted More generally, the CT-scan/finite- Gregory M. Erickson is in the Department of
throughout the Allosaurus skull during element modelling method could be applied Biological Science, Florida State University,
feeding. The overall result is an impressive to within- and between-species analyses. Conradi Building, Tallahassee, Florida 32306, USA.
biomechanical model of the skull’s form, Topics might include changes in physical e-mail: gerickson@bio.fsu.edu
function and performance. capacities during animal development and 1. Beaupré, G. S. & Carter, D. R. in Biomechanics: Structures
The model is extraordinary in its three- evolution; the effects of morphology and and Systems (ed. Biewener, A. A.) 149–174 (IRL, Oxford,
1992).
dimensional accuracy. Furthermore, it is scale on biomechanics; and the functional 2. Rayfield, E. J. et al. Nature 409, 1033–1037 (2001).
made at a regional (skull and jaw) anatom- significance of anatomical form in extinct 3. Hildebrand, S. F., Gilmore, C. W. & Cochran, D. M. Cold-
ical level, rather than a more common, animals (many of which cannot be modelled Blooded Vertebrates (Smithsonian Science Series, New York,
1930).
single-bone level. Detailed skull-and-jaw by living analogues). These exciting new 4. Laws, R. R. J. Vert. Paleont. 17, 59A–60A (1997).
finite-element analysis such as this has been areas of research require expertise from 5. Chapman, R. E. in The Complete Dinosaur (eds. Farlow, J. O.
attempted only rarely for living organisms, such disciplines as mechanical engineering, & Brett-Surman, M. K.) 112–135 (Indiana Univ. Press,
Bloomington, 1997).
for instance with the human models used in vertebrate palaeontology, evolutionary biol-
6. Keyak, J. H., Meagher, J. M., Skinner, H. B. & Mote, C. D. Jr
crash tests by the automobile industry. To ogy and comparative morphology. The J. Biomed. Eng. 12, 389–397 (1990).
date, the few applications in vertebrate trend towards an integrative approach in 7. Rensberger, J. M. in Functional Morphology in Vertebrate
palaeontology have been limited to descrip- many biology departments over the past Paleontology (ed. Thomason, J. J.) 151–172 (Cambridge Univ.
Press, 1995).
tions of single-bone mechanical loading in decade has encouraged a multidisciplinary 8. Carter, D. R., Mikic, M. & Padian, K. Zool. J. Linn. Soc. 123,
which stress distributions, not stress magni- approach11. Advances in computer technol- 163–178 (1998).
tude, were studied7,8. ogy, and the emergence of a new breed of 9. Thomason, J. J. Can J. Zool. 69, 2326–2333 (1991).
10. Erickson, G. M. et al. Nature 382, 706–708 (1996).
Rayfield and colleagues’ analysis indi- scientist versed in these various fields, point 11. Wake, M. H. Biol. Int. 39, 14–18 (2000).
cates that the maximal bite forces of the to a rosy future for biomechanical analyses of 12. Erickson, G. M. Proc. Natl Acad. Sci. USA 93, 14623–14627
1.4-tonne Allosaurus were surprisingly low, extinct organisms. s (1996).
on par with those for much smaller living
carnivorous mammals such as wolves and
leopards9, and at least six-times lower than Structural biology
sub-maximal estimates for its larger cousin,
Tyrannosaurus rex10. This by no means
suggests that the beast did not pack a lethal A 3D view of sodium channels
punch, only that it was discriminating in William A. Catterall
where and how it bit its prey. Rayfield et al.
contend that Allosaurus had a feeding strat- The voltage-gated sodium channel is essential for nerve cells to function.
egy reminiscent of that of Komodo dragons, But how does it work? A low-resolution, three-dimensional structure
inflicting rapid, slashing bites primarily to provides some provocative insights.
the soft tissues which meant that the prey
bled to death. Tyrannosaurus rex, on the
S
ensation, emotions, thought and particular functions of the sodium channel
other hand, with its much more robust teeth, movement all depend on voltage-gated (Fig. 1; reviewed in ref. 6). The positively
was capable of smashing through flesh and sodium channels — transmembrane charged fourth transmembrane segment
bone, and delivering deeper and perhaps proteins that initiate action potentials in (S4) in each domain serves as the voltage sen-
more instantly lethal bites10. nerve, muscle and other electrically excitable sor. The hydrophobic sixth transmembrane
Contrary to the general view that the cells. The hallmarks of sodium channels segments (S6) in the four domains form the
Allosaurus skull was delicate, the authors include rapid activation and inactivation wide inner lining of the pore. The preceding
argue that it was extraordinarily strong when the voltage across the plasma mem- SS1/SS2 segments (also known as P loops)
relative to the bite force — that is, it was brane of an excitable cell is depolarized form the narrower, ion-selective outer lining
‘overbuilt’. The strut-like skull structure was (voltage-dependent gating), and efficient of the pore. The short intracellular loop
mechanically well suited for dissipating and selective conduction of sodium ions connecting domains III and IV inactivates
biting forces and stresses. Rayfield et al. pro- through them. On page 1047 of this issue1, the sodium channel within one or two milli-
pose that the overbuilt skull was adapted Sato and colleagues provide the first three- seconds after it opens by folding over and
for absorbing forces from impacts with prey, dimensional view of the sodium-channel blocking the inner mouth of the pore. This
and from drawing its teeth through that prey. protein. The images reveal unexpected essential function is impaired in some inheri-
For the moment, however, it cannot be said structural features that may contribute to the ted diseases of hyperexcitability, including
whether this overbuilt skull structure reflects protein’s function in electrical excitability. periodic paralysis, cardiac arrhythmia and
these or other mechanical effects, non- The -subunit of the sodium channel, epilepsy (reviewed in ref. 7).
mechanical influences (such as accommo- which forms the ion-conducting pore, was What are the three-dimensional struc-
dation of soft tissues or sexual display), or initially identified as an unusually large tures of these functional elements of the
simply the typical reptilian condition. polypeptide (relative molecular mass, sodium channel? Initial insights came from
Rayfield and colleagues’ approach can be 260,000) by labelling with specific neuro- X-ray crystallography and nuclear magnetic
used to tackle other questions in dinosaur toxins2,3 that modify the generation of action resonance (NMR) analysis. The crystal
biology. Were overbuilt skulls common to all potentials. Subsequent DNA-cloning exper- structure of the pore region of a distantly
theropod dinosaurs? Which biomechanical iments revealed the amino-acid sequences related bacterial potassium channel8 defined
or other factors led to the development of of a family of nine related -subunits in a narrow outer pore formed by the P loops,
this and different skull morphologies? What mammals4,5. Each -subunit contains four and an inner pore surrounded by transmem-
was the mechanical significance of the homologous domains, I to IV, each with brane segments that are related to the S6
intracranial joints in dinosaur skulls? Fur- six predicted transmembrane segments segments of sodium channels. NMR analysis
ther loading tests with this model, as well as (Fig. 1). of the structure of the sodium channel’s
bite-force simulations using full-scale mod- Further work provided a two-dimen- inactivation gate in solution revealed a rigid
els of dinosaur teeth and jaws, might be used sional molecular map that showed which helical structure (an -helix), preceded by
to address some of these questions. structural features are responsible for which two turns that position a key hydrophobic
988 © 2001 Macmillan Magazines Ltd NATURE | VOL 409 | 22 FEBRUARY 2001 | www.nature.com

2. news and views
DAVID EHLERT
Top view
ΨΨ
I II III IV
Outside cell
– – – –
+ – + – + + +
12345 6 12345 6 1 234 5 6 123 45 6 Membrane
+ + + +
Inside cell h
Pore
Voltage sensing Inactivation
1 2
Amino Out
terminus Carboxy
terminus
In
3 4
Figure 1 Two-dimensional molecular map of the -subunit of the voltage-gated sodium channel,
Bottom view
showing how it folds and crosses the plasma membrane. The -subunit is composed of four
homologous domains (I to IV), each of which contains six transmembrane segments (1 to 6).
Cylinders represent -helical transmembrane segments. Other protein segments are illustrated as
straight lines, roughly in proportion to their length in the amino-acid sequence of the sodium
channel. Red indicates the S4 transmembrane segments, which serve as voltage sensors. Green
cylinders and shading indicate the S5 and S6 segments and the SS1/SS2 segment (P loop) between
them, which together form the walls of the sodium-ion-conducting pore. Circles indicate the
positions of amino acids (neutral, positively charged or negatively charged) that are important for
ion conductance and selectivity. indicates sites of N-linked glycosylation (that is, where a
carbohydrate group is attached). The circled ‘h’ (yellow) indicates the hydrophobic motif IFMT Cross-section
(isoleucine, phenylalanine, methionine, threonine) in the inactivation gate, which is thought to fold
1 2
into and block the ion-conducting pore.
amino-acid sequence (isoleucine, phenyl- by the large amino-terminal and carboxy-
alanine, methionine, threonine) so that it terminal regions and the large intracellular
can interact with and block the inner mouth loops connecting domains I, II and III. A
of the pore9. prominent indentation in one side of the
The structure of the complete sodium intracellular mass may hold the inactivation
channel1 now places these functional ele- gate formed by the short intracellular loop
ments in a three-dimensional context (Fig. connecting domains III and IV, as expected 3 4
2). Sato et al. determined this structure at 19 from the fact that this loop is accessible to
Å resolution by cryo-electron microscopy enzymes, antibodies and chemical reagents Figure 2 A three-dimensional representation of
and image reconstruction. The images show (reviewed in ref. 6). the voltage-gated sodium channel, based on
four transmembrane masses arrayed sym- The most unexpected features of the Sato et al.’s images1. Green indicates the central
metrically around a central axis that is per- structure are four peripherally located ion-conducting pore and its intracellular and
pendicular to the membrane. The central transmembrane pores, one in each domain extracellular connections to the bathing
axis is densely stained with ions from the (Fig. 2, red). What might these extra pores medium. Red indicates ‘gating’ pores, containing
bathing solution, providing an image of the do? During gating, the voltage sensors (the the S4 segments, and their connections to the
transmembrane pore and confirming the S4 segments) of a sodium channel are pre- central pore. Numbers 1 to 4 allow one to
widely accepted idea that the four domains dicted to move outwards across the mem- compare the orientations of the intact sodium
of the sodium channels are arranged roughly brane along a spiral path, thereby moving channel (second image from the top) and the
symmetrically around a central pore (Fig. 2, their positive gating charges across the cross-section (bottom). The dashed lines
green). Remarkably, the central pore does membrane’s electric field (reviewed in ref. show the position at which the cross-section
not connect directly to the intracellular and 6). Outward movement and rotation of the was taken.
extracellular bathing solutions. Instead, it S4 segments has indeed been detected with
splits into four branches that connect it to specific chemical labelling methods10–12. Surprisingly, these apparent gating pores
the bathing solutions at each end (Fig. 2, The large outward movements observed for are not linear, even though they are thought
green). these transmembrane helices led to the pro- to accommodate the movement of a linear
As viewed parallel to the membrane posal that these segments move through S4 segment across the membrane. Perhaps
surface, the sodium-channel protein is bell- special, narrow-waisted gating pores in the structure observed in these images is a
shaped, with about 47% of its mass on the each domain, designed to allow the move- closed state in which transmembrane move-
intracellular side of the membrane and 24% ment of substantial gating charge across ment of the S4 segments is prevented. If so,
on the extracellular side (Fig. 2). These the membrane with minimal physical voltage-dependent activation of the channel
observations are in general agreement with motion10. The four narrow transmembrane would require a conformational change in
expectations from two-dimensional models pores now observed in each domain of the the gating pores to allow movement of the
of the protein structure (Fig. 1). Much of structure are prime candidates for these gating charge, as well as an outward trans-
the intracellular mass must be contributed gating pores. location and rotation of the S4 segments
NATURE | VOL 409 | 22 FEBRUARY 2001 | www.nature.com © 2001 Macmillan Magazines Ltd 989

3. news and views
themselves. This would be a previously William A. Catterall is in the Department of tion). TD is tidal dissipation, which is related
unrecognized step in the gating process. Pharmacology, School of Medicine, University of to the exchange of angular momentum
This view of the sodium channel at 19 Å Washington, Seattle, Washington 98195-7280, USA. between the Earth and the Moon, and results
resolution whets our appetite for higher- e-mail: wcatt@u.washington.edu in changes in the Earth–Moon distance, in
resolution images. Analyses of such images, 1. Sato, C. et al. Nature 409, 1047–1051 (2001). the spin velocity of the Earth, and in the
together with further structure-based studies 2. Beneski, D. A. & Catterall, W. A. Proc. Natl Acad. Sci. USA 77, velocity of the Moon’s rotation about the
639–642 (1980).
of the channel’s function, will help to answer 3. Agnew, W. A., Moore, A. C., Levinson, S. R. & Raftery, M. A. Earth. Both dynamical ellipticity and tidal
many questions. For example,which mol- Biochem. Biophys. Res. Commun. 92, 860–866 (1980). dissipation affect Earth’s obliquity and pre-
ecular interactions underlie the gating pro- 4. Noda, M. et al. Nature 312, 121–127 (1984). cession, and have to be determined indepen-
5. Goldin, A. L. et al. Neuron 28, 365–368 (2000).
cess? How do the gating pores observed here 6. Catterall, W. A. Neuron 26, 13–25 (2000).
dently of astronomical theory. Each pair of
participate in the overall regulation of sodi- 7. Lehmann-Horn, F. & Jurkat-Rott, K. Physiol. Rev. 79, parameters gives rise to one solution. Faced
um-channel function? And how does the 1317–1372 (1999). with a large number of slightly different
8. Doyle, D. A. et al. Science 280, 69–77 (1998).
structure of the central pore allow the selec- 9. Rohl, C. A., Boeckman, F. A., Scheuer, T., Catterall, W. A. &
solutions, it becomes necessary to try to
tive passage of sodium ions? In the meantime, Klevit, R. Biochemistry 38, 855–861 (1999). identify the ‘best’ one.
we can celebrate this latest achievement, 10. Yang, N., George, A. L. & Horn, R. Neuron 16, 113–122 (1996). In one earlier attempt, Lourens et al.5
11. Cha, A., Snyder, G. E., Selvin, P. R. & Bezanilla, F. Nature 402,
which provides new insight into the struc- correlated cycles seen in Mediterranean
809–813 (1999).
tural basis of voltage-dependent gating, and 12. Glauner, K. S., Manuzzu, L. M., Gandhi, C. S. & Isacoff, E. sediments over the past five million years
holds great promise for the future. s Nature 402, 813–817 (1999). with summer insolation derived from differ-
ent astronomical solutions, one of which was
La90(1,1) — that is, with the present-day value
Earth history for dynamical ellipticity ( 0.9 would mean
0.9 times today’s value) and the present-day
Sediments to planetary motion value for tidal friction (TD 0 would mean
no tidal dissipation). They looked at inter-
Marie-France Loutre ference patterns between precession and
obliquity that were recognizable in both the
Celestial mechanics has long been used as an aid for interpreting sediment data and computed insolation. A
geological records. The compliment is being repaid through analysis of cross-spectral comparison between the data
Mediterranean sediments. and summer insolation time series led them
to conclude that the La90(1,1) is the most accu-
(the planet’s varying position on its orbit at
T
heories about the variation in Earth’s rate astronomical solution from a geological
orbit have proved helpful in setting the the beginning of the seasons). point of view — in other words, it results in
chronology for geological records of There are of course limitations to this tun- the best fit with the geological record over the
climate. But the reverse can also be true — ing technique. The climate cycles in the data past five million years.
such records can aid astronomers in com- series must be recognized a priori, proceeding More recently, Pälike and Shackleton6
puting Earth’s orbital parameters at times in from causes to effects, and the climate record applied a method based on interference pat-
the remote past. That is the aim of Lourens et must contain at least some independent indi- terns between the precession and obliquity
al., as they describe on page 1029 of this cation of its age (a so-called control point). components derived from geological data,
issue1. They have taken advantage of an The lag between the astronomically induced and the astronomical solutions, to identify
exceptional sediment record, which can be change and the climate response is usually small changes in tidal dissipation and
related to insolation (the amount of solar assumed to be constant, although it may be dynamical ellipticity. They showed that these
radiation reaching Earth’s upper atmos- different in (for instance) glacial and inter- parameters are likely to have remained close
phere). They come to certain conclusions glacial times. Finally, the accuracy of the to the present-day values for the past 25
about the planet’s rotational history over the chosen target curve — that is, the computed million years.
past three million years, and create an astro- orbital parameters, insolation, or both — In their new work1, Lourens et al. present
nomical timescale to be used in further is limited, in particular by our imperfect a comprehensive study to estimate dynami-
research. knowledge of Solar System dynamics. cal ellipticity and tidal dissipation in the
Climate records for all except the very As climate records became longer and La90( ,TD) solution between 2.5 and 3 million
recent past exist only as proxies, not as direct older, it became increasingly necessary to years ago. Their work is made possible by a
measurements. Such proxies are derived have highly accurate time series showing past high-resolution record of the ratio of titani-
from tree rings or cores from various loca- variation in orbital parameters. Hence the um to aluminium from a sediment core in
tions — the deep seas, lakes, continents, ice proposal at a conference2 some years ago that the eastern Mediterranean Sea. For compli-
sheets or glaciers. The timescale is usually the geological record could be used to obtain cated reasons to do with the balance between
drawn from radioactive-decay techniques, estimates of past variation in orbital parame- airborne and river input of material into the
allowing the development of a good chron- ters, and that those estimates could be used Mediterranean, this ratio provides a good
ology for up to a few hundred thousand to constrain the predictions based on celes- estimate of humidity over northern Africa,
years ago. Another procedure has also been tial mechanics . which is controlled by summer insolation.
devised to match (tune) the features in the By the early 1990s, slightly different The data series is tuned against an astronom-
proxy records to those of the Earth’s orbital astronomical solutions became available for ical index computed for different La90( ,TD)
parameters or of the astronomically derived the period up to about several million years astronomical solutions — that is, the astro-
insolation (or both). Since the 1980s it has ago (see ref. 3 for review). One of them, nomical ages of the target are assigned to cor-
become clear that much climate variation known as the La90( ,TD) solution4, accounts relative peaks in the data series. The highest
stems in some way from changes in insol- for two factors that influence Earth’s orbit. linear correlations (the optimal fits) between
ation driven by various astronomical factors The first, , is the dynamical ellipticity, the target and the data series are obtained
— Earth’s eccentricity (the shape of its orbit which is affected by changes in the distribu- when the target is based on solutions ranging
around the Sun), its obliquity (inclination tion of masses on and in the Earth (ice caps, between La90(1.0003,0) and La90(0.9997,1). Such a
of the axis of rotation), and its precession for instance, or patterns of mantle convec- high correlation is not unequivocal evidence
NATURE | VOL 409 | 22 FEBRUARY 2001 | www.nature.com © 2001 Macmillan Magazines Ltd 991