Download Virus on virus infects bacterium

news and views
between spin ice and water ice. This work
reveals that two analogous systems —
Dy2Ti2O7 and ice — have the same entropy,
even though the ordering dynamics of the
spins and hydrogen atoms are vastly different. Yet these two materials still have the
same statistical behaviour, pointing to a
deeper truth behind Pauling’s finite entropy.
It is easy to believe that ice itself might be out
of equilibrium, but it is harder to believe the
same of Dy2Ti2O7, because there is, as yet, no
evidence of a transition into a glassy state like
that of ice. One might conclude that spin ice
presents a real threat to the third law of thermodynamics. That would be premature, but
it certainly represents plausible evidence that
the disordered state of water ice is the real
ground state at low temperatures.
It would be no exaggeration to say that
such a transparent and easily interpretable
result has never before been obtained from a
frustrated magnet. This highlights what is so
appealing to the experimentalist about spinice compounds: despite the inevitable presence of defects and other minor perturba-
tions that frequently dominate other frustrated magnets, an inherent purity of behaviour comes through. The link between theory and experiment is therefore much more
straightforward. It is clear that, because of
this, spin ice has the potential to illuminate
other vexing questions arising in the past 30
years. The most urgent is the origin of glassy
behaviour in chemically ordered magnets, a
question that reaches into the heart of what
causes glassiness in the first place.
Mark Harris is at the ISIS Facility, Rutherford
Appleton Laboratory, Chilton, Didcot,
Oxfordshire OX11 0QX, UK.
e-mail: [email protected]
1. Anderson, P. W. Science 235, 1196–1198 (1986).
2. Ramirez, A. P., Hayashi, A., Cava, R. J., Siddharthan, R. &
Shastry, B. S. Nature 399, 333–335 (1999).
3. Harris, M. J., Bramwell, S. T., McMorrow, D. F., Zeiske, T. &
Godfrey, K. W. Phys. Rev. Lett. 79, 2554–2557 (1997).
4. Moessner, R. Phys. Rev. B 57, R5587–R5589 (1998).
5. Bramwell, S. T. & Harris, M. J. J. Phys. Condens. Matt. 10,
L215–L220 (1998).
6. Pauling, L. J. Am. Chem. Soc. 57, 2680–2684 (1935).
7. Giauque, W. F. & Stout, J. W. J. Am. Chem. Soc. 58, 1144–1150
(1936).
8. Blöte, H. W. J., Wielinga, R. F. & Huiskamp, W. J. Physica 43,
549–568 (1969).
Microbiology
Virus on virus infects bacterium
Ronald K. Taylor
any bacteria carry segments of DNA
called pathogenicity islands, which
are enriched for genes encoding the
proteins that contribute to virulence. Such
proteins include secreted toxins, factors and
signalling devices that help the bacterium to
colonize its host, and even specialized types
of molecular apparatus that inject bacterial
products into host cells. We are starting to
understand how the bacteria acquired these
pathogenicity islands during evolution, and
the latest findings are reported by Karaolis et
al.1 on page 375 of this issue. They describe
a fascinating interplay between two pathogenicity islands — the first allowed the bacterium Vibrio cholerae to be infected with the
second, which is a bacterial virus (bacteriophage) that carries the cholera-toxin genes.
Over 100 years ago, V. cholerae was
identified as the bacterial species that causes
the deadly, infectious diarrhoeal disease
cholera2. More recently, the potent cholera
toxin was found to be encoded by a filamentous bacteriophage called cholera-toxin
phage (CTXF), which exists within V.
cholerae 3. The CTXF contains singlestranded DNA, and it uses the bacterium’s
main colonization pilus — an appendage
used to infect the host’s intestine — as its
receptor4,5. This pilus, termed toxin co-regulated pilus (TCP), is a fibre of polymerized
pilin protein termed TcpA. It promotes
colonization by enhancing interactions
between bacterial cells and, perhaps, directly
with the host.
M
312
The TCP belongs to the type-4 family of
pili, members of which are involved in
colonizing diverse bacterial species6. As the
name implies, expression of the genes that
encode the TCP parallels expression of the
tcpPH toxT
toxin genes7 — that is, they are expressed
only when the bacterium finds itself in a
suitable environment such as the epithelial
surface of the intestine. The genes encoding
the structural and assembly components of
the TCP are located on a pathogenicity island
known as the V. cholerae pathogenicity
island (VPI)8. All strains of V. cholerae that
can cause epidemic cholera contain a VPI,
and it is thought to be transferred between
strains in the environment. Certain features
of the VPI and its chromosomal location
suggest that this genetic element might be
encoded by a bacteriophage1,8,9.
Karaolis and colleagues1 now show that
the VPI is, indeed, a filamentous bacteriophage. Termed the VPI phage (VPIF), it can
transmit itself between certain strains of V.
cholerae, and it carries ssDNA in a similar
way to CTXF. The protein that surrounds
the DNA (the coat protein) is TcpA, a finding
that has several astounding implications.
First and foremost, it implies that the same
structure may be both a colonization pilus
and a bacteriophage particle. The authors
propose that the bacteriophage can be
formed and bud off from the bacterium
while acting as a colonization pilus. If this is
true, it indicates that the VPIF is the receptor
for yet another phage — the CTXF (Fig. 1).
But it is a mystery how an infectious phage
particle, which must be released from the
bacterial cell to infect another cell, can also
act as a colonization factor and a phage
receptor.
One possibility is that some of the TcpA
produced by the bacterium serves one func-
toxRS
ctxAB
ompT ompU
T
U
CTXΦ
VPIΦ
Toxin
TCP
Figure 1 Regulation of virulence genes in pathogenic Vibrio cholerae. The V. cholerae pathogenicity
island phage (VPIF) interacts with a recipient V. cholerae cell, resulting in infection and insertion of
VPIF DNA into the bacterial chromosome. This allows production of the toxin co-regulated pili
(TCP), which, in turn, allows infection by the cholera-toxin phage (CTXF). The bacterium now has a
full complement of virulence genes. Their expression is controlled by complex interactions between
regulators encoded within the ancestral chromosome and those encoded on the newly acquired
genetic elements. For example, the toxR gene product activates the expression of other genes
encoding proteins in the bacterium’s outer membrane (such as ompU in V. cholerae). In concert with
the tcpPH genes (which are encoded by the VPI), toxR also activates expression of the VPI-encoded
regulatory toxT gene. The ToxT protein (T) is required for expression of VPI genes, such as the tcp
genes and the ctx operon of the CTXF, but it does not regulate expression of ompU 7,11.
© 1999 Macmillan Magazines Ltd
NATURE | VOL 399 | 27 MAY 1999 | www.nature.com
news and views
tion, with the remaining TcpA serving the
other. For example, not all strains of V.
cholerae that express TCP can form transferable bacteriophage particles. Perhaps most
of the TcpA is incorporated into a colonization pilus, and the remainder is used to coat
the DNA in infectious phage particles (but
only in those strains that can produce transferable phage). Interestingly, those strains
that do not make accompanying phage were
identified long before the strains that do.
Maybe the bacteriophage has been caught in
different stages of evolution — the newer
strains can still produce the phage, whereas
the more established strains have been
selected for their ability to colonize host cells
rather than for their capacity to transfer
genes on a phage.
The VPIF is unusual in other ways. It is
very large for a ssDNA filamentous phage,
and the replicative form of its DNA is not
very abundant. These properties may be
related to the many functions of the genes
that it encodes, or they may reflect a regulatory mechanism that balances development
of the phage with formation of the pilus.
Another amazing link between the VPIF
and CTXF is the intimate co-regulation of
gene expression. As shown in Fig. 1, virulence of V. cholerae involves a fascinating
interplay between genes and regulatory elements encoded by the bacterial chromosome
and those on the VPIF and CTXF. The fact
that the TCP serves as the CTXF receptor
indicates that, during evolution of toxinproducing strains such as V. cholerae, the
VPIF has to be acquired before the CTXF
can infect the bacterial strain. This sequence
of infection ensures that the regulatory
controls for interaction with CTXF genes
are already in place.
The extent to which type-4 pili and bacteriophages are intertwined in other bacterial
species remains to be seen. Many type-4 pili
have other features, such as twitching motility, which may be involved in colonization or
even in bacterial dispersal. But such functions would probably not be involved in
phage production. In many cases, the genes
that encode bacterial pili are not associated
with a particular pathogenicity island. So,
perhaps further links should first be pursued
with those genes that are associated with
pathogenicity islands. Whatever the outcome of such investigations, bacteriophages
will continue to surprise us with their multifaceted roles in the evolution of pathogenic
and other bacterial species10.
Ronald K. Taylor is in the Department of
Microbiology, Dartmouth Medical School, Hanover,
New Hampshire 03755, USA.
e-mail: [email protected]
1. Karaolis, D. K. R., Somara, S., Maneval, D. R. Jr, Johnson, J. A.
& Kaper, J. B. Nature 399, 375–379 (1999).
2. Kaper, J. B., Morris, J. G. & Levine, M. M. Clin. Microbiol. Rev.
8, 48–86 (1995).
3. Waldor, M. K. et al. Mol. Microbiol. 24, 917–926
(1997).
NATURE | VOL 399 | 27 MAY 1999 | www.nature.com
4. Taylor, R. K. et al. Proc. Natl Acad. Sci. USA 84, 2833–2837
(1987).
5. Tacket, C. O. et al. Infect. Immunol. 66, 692–695 (1998).
6. Strom, M. S. & Lory, S. Annu. Rev. Microbiol. 47, 565–596
(1993).
7. Skorupski, K. & Taylor, R. K. Mol. Microbiol. 25, 1003–1009
(1997).
8. Karaolis, D. K. R. et al. Proc. Natl Acad. Sci. USA 95, 3134–3139
(1998).
9. Kovach, M., Shaffer, M. & Peterson, K. Microbiology 142,
2165–2174 (1996).
10. Waldor, M. K. Trends Microbiol. 6, 295–297 (1998).
11. Häse, C. C. & Mekalanos, J. J. Proc. Natl Acad. Sci. USA 95,
730–734 (1998).
Palaeoclimatology
Warming without high CO2?
Benjamin P. Flower
ver the next century, increasing levels
of greenhouse gases are expected to
raise Earth’s mean surface temperatures by some 2–5 7C. To help understand
future climate change, Earth scientists are
examining ancient periods of extreme
warmth such as the Miocene Climatic Optimum of about 14.5–17 million years ago
(Myr). Fossil floral and faunal evidence indicate that this was the warmest time of the past
35 million years; mid-latitude temperatures
were as much as 6 7C higher than at present.
Many workers believe that high CO2 levels, in
combination with oceanographic changes,
caused Miocene global warming by the
greenhouse effect. But in the June issue of
Paleoceanography, Pagani, Arthur and Freeman1 present evidence for surprisingly low
CO2 levels of about 180–290 parts per million
by volume (p.p.m.v.) throughout the early to
late Miocene (9–25 Myr). These authors conclude that greenhouse warming by high CO2
cannot explain Miocene warmth, and that
other mechanisms must have had a greater
influence.
In their study, Pagani et al. used a relatively
new molecular-isotopic approach that has
been effective in tracing CO2 in glacial–interglacial cycles of the past several hundred
thousand years2,3. A record of surface water
[CO2(aq)] can be obtained from C37:2 alkenones in marine sediments, because isotopic
fractionation during phytoplankton photosynthesis is largely controlled by [CO2(aq)],
cell growth rate and cell geometry. Assuming
the effects of cell growth rate and cell geometry are minimal in low-nutrient waters,
[CO2(aq)] may be estimated by comparing
C37:2 alkenone d13C with the d13C of total
CO2 as recorded by planktonic foraminifera.
Atmospheric CO2 can then be calculated
from [CO2(aq)] values based on Henry’s law.
Pagani and colleagues’ CO2 record bears
little relation to an expected association
between CO2 and long-term climate change
during the Miocene (Fig. 1, overleaf). In
particular, low CO2 values span the warm
Climatic Optimum which ended around
14.5 Myr, whereas rising CO2 levels accompany global cooling and expansion of the
East Antarctic Ice Sheet about 12.5–14 Myr
(as inferred from the oxygen-isotope
composition of benthic foraminifera from
deep-sea sediment cores). Rising CO2 during
O
© 1999 Macmillan Magazines Ltd
the middle Miocene may eliminate a reverse
greenhouse effect as the main cause of East
Antarctic Ice Sheet expansion.
Several sources of uncertainty in calculating [CO2(aq)] may increase CO2 estimates,
but only minimally. Miocene [CO2(aq)] values
may be artificially low if the d13C of surfacewater total CO2 is overestimated. Foraminiferal d13C may overestimate [CO2(aq)] either
because of species-dependent effects or
because of the carbonate-ion effect on seawater d13C (ref. 4). However, [CO2(aq)] values
are primarily controlled by variations in the
d13C of C37:2 alkenones, and even a 0.5–1‰
overestimation of surface-water d13C would
yield only slightly higher CO2 values1.
Similarly, underestimation of sea-surface
temperature from foraminiferal d18O (due to
a possibly lesser ice-volume effect) would
increase calculated CO2 only slightly.
Pagani and colleagues’ low CO2 values of
180–290 p.p.m.v. (present-day value is 360
p.p.m.v.) will fuel debate on long-term CO2
history during the Cenozoic era. Some geochemical models5–8 suggest that increased
global chemical weathering and erosion during the past 40 Myr progressively consumed
atmospheric CO2. Others argue that there
was little change in either global weathering
rates or CO2 (ref. 9). Low Miocene CO2 levels
are consistent with palaeo-pH data10, which
indicate little difference from the presentday ocean. Similarly, CO2 trends on the
timescale of 105 –106 years are in line with
foraminiferal d13C records of organiccarbon burial relative to calcium carbonate
(Fig. 1). High d13C values between 13 and
17.5 Myr, termed the Monterey Excursion,
indicate greater burial of organic carbon
which may have been sufficient to draw
down CO2 from the atmosphere during this
time11. Thus, low Miocene CO2 is supported
by some independent geochemical data.
Furthermore, Pagani et al. find that shortterm CO2 minima coincide with glacial cooling on the 104–105-year scale. Coeval maxima
in foraminiferal d13C support the idea that
organic-carbon burial was linked to CO2
drawdown on such timescales12,13. This
finding squares with known mechanisms of
climatic change during the past several
hundred thousand years, in which high CO2
clearly increased global warming. In short,
CO2 variability seems to contribute to
313