14 JULY 2017 • VOL 357 ISSUE 6347 129 SCIENCE sciencemag.org
We should not, however, disregard Gilar-
ranz et al.’s other findings. As the authors
show, modularity does not necessarily result
in the best outcome in all situations. The em-
pirical system that the authors analyzed pro-
vides a very good proxy for other networked
systems in which nodes receive, produce,
and exchange flows. Consider, for instance,
people exchanging information through on-
line platforms. A strong modular component
can result in the appearance of bottlenecks,
thus preventing the efficient spread of infor-
mation through the network. This might be
remedied by introducing long-range short-
cuts into the system, but these modifications
would not be smart strategies for containing
the spread of malicious information.
A more complex networked system is the
interbank payment network, which has been
in the spotlight since the 2008 financial crisis. It has been argued that the interconnectedness of economies, markets, information
flow, and disruptive events such as natural
disasters, the refugee crisis in Europe, or
Brexit makes the financial system extremely
vulnerable, increasing the risk of small perturbations resulting in severe consequences
(3, 12–14). The question that the scientific
community is trying to answer in this context
is how a local perturbation became so amplified and what strategies might prevent further worldwide crises (14, 15).
It may be possible to use principles of
natural design to reshape the organization
of the financial system and mitigate these
risks (3, 14, 15). However, it remains to be
explored how the concept of modularity can
be exported to interconnected networked
systems. Connecting modular networked
systems does not ensure modularity on a
global scale. Additionally, the nature of the
connections between networked systems
will likely play a critical role in their capacity to contain perturbations. Hopefully, the
study of natural systems will soon provide
us with clues about how to design robust
interconnected networked systems. j
1. R.M.May, Nature 238,413(1972).
2. R. M. May etal. , Nature 451, 893 (2008).
3. D.Helbing, Nature 497,51(2013).
4. J.Grilli, T.Rogers, S.Allesina, Nat.Commun. 7,12031(2016).
5. L.J.Gilarranz etal., Science 357, 199(2017).
6. R. Guimerà etal., Ecology 91,2941(2010).
7. M. E. J. Newman, M. Girvan, Phys. Rev. E 69, 026113 (2004).
8. R. Guimerà, M. Sales-Pardo, L. A. N. Amaral, Phys. Rev. E 70,
9. R.Guimerà, M.Sales-Pardo,L.A.N. Amaral, Nat. Phys. 3,63
10. S.Fortunato, Phys. Rep. 486,75(2010).
11. U.Alon, Science 301,1866(2003).
12. A.Vespignani, Nature 464,984(2010).
13. J. Gao, S. V. Buldyrev, H. E. Stanley, S. Havlin, Nat. Phys. 8, 40
14. S.A. Levin,A. W.Lo, Proc. Natl. Acad. Sci. U.S. A. 112, 12543
15. M.Bardoscia, S.Battiston, F.Caccioli, G.Caldarelli, Nat.
Commun. 8, 14416 (2017).
Immunology taught by rats
A rodent model of hepatitis C virus infection should
guide therapeutics and vaccines
By Paul Klenerman and Eleanor J. Barnes
Immunology may be best taught by vi- ruses (1), and possibly by humans (2), but the rats of New York City surprisingly also have plenty to offer. A survey pub- lished in 2014 of the pathogens carried by rats trapped in houses and parks in
Manhattan identified a huge burden of infectious agents in these animals, including
several novel viruses (3). Among these are
Norway rat hepaciviruses (NrHVs), which
belong to the same family as hepatitis C virus (HCV). NrHVs were found in rat livers,
raising the possibility of establishing a small
animal model of human HCV infection. On
page 204 of this issue, Billerbeck et al. (4)
fulfill this prediction.
HCV is a major human pathogen and part
of a viral family that has recently expanded to
include viruses in horses, dogs, and deer mice
(5). It causes persistent liver inflammation
and, with time, leads to cirrhosis and possibly
liver cancer. For many years, treatments were
based on the cytokine interferon-a (which
boosts host antiviral responses), but this had
many side effects and cured only 50 to 70%
of patients. The advent of a range of new oral
combined-drug regimens [targeting HCV
protease, nonstructural protein 5A (NS5A),
and polymerase] has transformed the field,
with a cure for HCV achievable in more than
95% of patients in many different groups (6).
Despite this remarkable achievement in
therapy, there are many unanswered ques-
tions about HCV. These include the role of
the host immune response in controlling
the infection and the potential for vaccina-
tion. After infection, about 25% of individu-
als clear HCV through innate and adaptive
immunity, notably CD4+ T cell and CD8+ T
cell responses. In most patients, however, the
virus persists long term, evading and sub-
verting these responses. Because the early
stages of HCV infection are often clinically
silent and the main site of immunological
activity—the liver—is not readily accessible,
the early innate and adaptive responses as-
sociated with controlling HCV have not been
well defined. Hence, the development of a
small animal model could help address both
the fundamental features of a hepatotropic
virus and vaccine strategies that are based on
priming effective host immunity.
Billerbeck et al. set out to establish whether
they could infect immunocompromised and
immunocompetent mouse strains with NrHV.
Mice lacking type I interferon signaling and
adaptive immunity readily became persistently infected. Immunocompetent animals
cleared the virus over a few weeks; the
length of time depended on the animal’s age
and whether the virus adapted to the mouse
through serial passage. The mutations that facilitated adaptation were mainly in the viral
envelope glycoproteins, suggesting that viral
entry into host cells or antibody binding to the
virus had become modified.
Having established the mouse model of
NrHV infection, Billerbeck et al. explored the
immune factors controlling the virus. The
authors identified specific T cells that infiltrated the liver following acute infection and
found roles for CD4+ T cells and CD8+ T cells
that appeared similar to those observed during human HCV infection and in response to
vaccination of chimpanzees (7, 8). Interestingly, even transient early depletion of CD4+
T helper cells in the mouse model of NrHV
infection resulted in viral persistence over
several months, associated with immune
exhaustion. This places the T helper cell response at the center of immune control, in
agreement with human genetic and experimental data (e.g., ex vivo T cell proliferation
and cytokine responses) (9).
The impact of transient CD4+ T cell depletion is also reminiscent of observations from
the well-established lymphocytic choriomeningitis virus (LCMV) mouse model of viral
persistence. A specific LCMV strain (clone
13) can slowly be cleared from the blood and
liver of an infected mouse over many weeks,
but short-term loss of CD4+ T helper cells can
“…NrHV may have
to teach us about viral
control of infection…”
Translational Gastroenterology Unit and Peter Medawar
Building for Pathogen Research, Nuffield Department of
Medicine, University of Oxford, Oxford OX1 3SY, UK.