(130f) Controllability of the Influenza Virus-Host Protein-Protein Interaction Network: Engineering Insights into Host-Virus Interactions

Controllability of the Influenza
Virus-Host Protein-Protein Interaction Network: Engineering Insights into
Host-Virus Interactions

Emily Ackerman, University of
Pittsburgh

Influenza virus is a major threat
to public health each year. The World Health Organization estimates that there
are 3-5 million severe cases of seasonal influenza worldwide yearly, with
290,000-650,000 cases resulting in death¹. Yearly seasonal outbreaks are
especially harmful to at-risk populations such as children and pregnant women,
the elderly, and the immunocompromised^2,3. Additionally, there is a
continual threat of a pandemic, with the most recent being the 2009 H1N1 outbreak.
As such, there is a high demand for effective treatment strategies and novel
methods for antiviral drug target discovery.

Understanding which cellular
components are key in enabling viruses to interrupt the normal activity of
healthy cells provides a major advantage in the design of novel disease
treatments^4,5. Systems biology methods aim to
investigate this problem through the application of mathematical and
computational based methods. While kinetic modeling approaches detail and
quantify cellular events in relevant pathways, the effectiveness of the
approach can be limited by data availability and lack of interaction mechanism
knowledge. Alternatively, protein-protein interaction (PPI) data is used to
explore the same biological information in a convenient, highly simplified
network representation⁶. Though the method lacks the
quantitative nature of ODE modeling, the generality of cell interactions
opposed to specific pathway analysis allows for the study of PPI relationships on
a whole cell level. Previous work with PPI networks have identified
disease-associated genes and predicted drug targets using established network
topology (or centrality) measures such as degree and betweenness^7–9. However, there is limited work
describing the ways in which PPI networks are differently regulated when
comparing changing cell states.

In classic control theory,
controllability is the idea that a system can be driven to any final state in
state space given an appropriate external input in finite time¹⁰. A PPI network represents a cell
system of protein states. During infection, viral proteins manipulate host
proteins to control the system and drive it to a final infected state which performs
virus replication tasks. While PPI systems are too large to directly calculate
the controllability matrix, two methods of network controllability analysis exist
which use bipartite graphs to simplify the associated calculations. One
analysis determines if it is more or less difficult to control the network in
the absence of each protein¹¹. Ultimately, this analysis
calculates the minimum set of nodes needed to control the network with or
without a selected protein. This set is non-unique; which leads to the second
analysis which places each protein in the broad context of its importance to all
configurations of network control¹². Prior to this work, only the
first method has been applied to a human PPI network¹³. These analyses identify key
proteins being manipulated in the externally controlled, infected cell system.

Here, we complete a dual
controllability analysis of two PPI networks: a host PPI network (HIN) and an
integrated virus-host PPI network (VIN). This design allows for the comparison
of changes between the healthy and infected cell states to identify proteins
that display changes to the regulatory behavior of the network. Further, the
analysis compares the characteristics of influenza A virus (IAV) interacting
host proteins and “driver” proteins (proteins which must be manipulated for the
system to be fully controlled, a non-unique set) to all proteins of the network
to determine if these specialized groups are enriched for virus replication
host factors; i.e. human proteins required for influenza virus replication.

An observed increase in the basic
topological measurements of degree and betweenness after the addition of viral
interactions to the network is proof of the wide-reaching effects to cellular
behavior during infection. This effect is especially strong for proteins that
are both IAV interacting and driver proteins. The controllability analyses determine
that it is more difficult to control the overall network in the absence of IAV
interacting host proteins, and that they are well integrated into the control
structure without requiring their own manipulation for total network control.
This suggests that viruses prefer to interact with host proteins that offer the
best advantages in controlling total system behavior. Conversely,
controllability results predict that the absence of driver proteins makes it
easier to control the network, meaning that they act as a barrier to control of
the system. Over 75% of driver proteins are able to act as non-driver nodes
given certain control configurations, opening up the idea of viral escape
routes: the idea that virus proteins could use alternative pathways under
pressure and maintain system control.

The first controllability
classification does not identify proteins of interest in the contrast between
the healthy HIN and infected VIN, despite known changes to the cell system
during infection. However, the second controllability classification identifies
a set of 24 proteins which exhibit notable changes between the HIN and VIN
systems. Together, they are predicted as indicators of infected state
regulatory roles, with the most prominent being protein arginine
methyltransferase 5 (Entrez ID: 10419) with a VIN betweenness of 250 times its
HIN value. 50% of the 24 proteins are identified as having interferon
regulating roles and an IPA¹⁴ analysis shows that they are
centered around NF-kB, which has been previously identified as inhibited by
influenza virus^15,16. Though the set of 24 proteins are
not significantly enriched in the results of six siRNA screening studies for
host factors of influenza A replication, one protein, heterogeneous nuclear
riboprotein A0 (Entrez ID: 10949), is triple validated. This rare observation suggests
that controllability methods hold power for the prediction of disease host
factors.

By analyzing the control structure
of the cell system, we are able to predict of the protein components which
allow an influenza A infection to take hold. Disruption of these methods of
viral control may prove to be an effective way of disrupting the infection as a
whole. As such, the identification of these key disease proteins advances the
prediction of possible antiviral drug targets. Additionally, this approach
enables the study of alternative pathways for maintaining viral replication
which could lead to drug resistance. A window into the control behavior and robustness
of influenza infection could position us one step ahead of both yearly,
seasonal outbreaks as well as the potential of pandemic infection.

1. WHO.
Influenza (Seasonal) Fact Sheet. (2017).

2. Mertz, D. et
al. Populations at risk for severe or complicated influenza illness:
systematic review and meta-analysis. BMJ 347, f5061--f5061
(2013).

3. Rodríguez-Baño,
J., Paño-Pardo, J. R., Múñez Rubio, E. & Segura Porta, F. Pregnancy,
obesity and other risk factors for complications in influenza A(H1N1) pdm09
infection. Enferm. Infecc. Microbiol. Clin. 30, 32–37 (2012).

4. Klipp, E.
& Liebermeister, W. Mathematical modeling of intracellular signaling
pathways. BMC Neuroscience 7, (2006).

5. Schoeberl,
B., Eichler-Jonsson, C., Gilles, E. D. & Muüller, G. Computational modeling
of the dynamics of the MAP kinase cascade activated by surface and internalized
EGF receptors. Nat. Biotechnol. 20, 370–375 (2002).

6. Cho, D.-Y.,
Kim, Y.-A. & Przytycka, T. M. Chapter 5: Network Biology Approach to
Complex Diseases. PLoS Comput. Biol. 8, e1002820 (2012).

7. Jonsson, P.
F. & Bates, P. A. Global topological features of cancer proteins in the
human interactome. Bioinformatics 22, 2291–2297 (2006).

8. Hase, T.,
Tanaka, H., Suzuki, Y., Nakagawa, S. & Kitano, H. Structure of protein
interaction networks and their implications on drug design. PLoS Comput.
Biol. 5, (2009).

9. Yildirim, M.
a, Goh, K.-I., Cusick, M. E., Barabási, A.-L. & Vidal, M. Drug-target
network. Nat. Biotechnol. 25, 1119–26 (2007).

10. Lin, C. T.
Structural Controllability. IEEE Trans. Automat. Contr. 19,
201–208 (1974).

11. Liu, Y. Y.,
Slotine, J. J. & Barabási, A. L. Controllability of complex networks. Nature
473, 167–173 (2011).

12. Jia, T. &
Barabási, A. L. Control capacity and a random sampling method in exploring
controllability of complex networks. Sci. Rep. 3, (2013).

13. Vinayagam, A. et
al. Controllability analysis of the directed human protein interaction
network identifies disease genes and drug targets. Proc. Natl. Acad. Sci.
113, 4976–4981 (2016).

14. Krämer, A.,
Green, J., Pollard, J. & Tugendreich, S. Causal analysis approaches in
ingenuity pathway analysis. Bioinformatics 30, 523–530 (2014).

15. Kumar, N.,
Xin, Z.-T., Liang, Y., Ly, H. & Liang, Y. NF-kappaB signaling
differentially regulates influenza virus RNA synthesis. J. Virol. 82,
9880–9 (2008).

16. Ludwig, S.
& Planz, O. Influenza viruses and the NF-κB signaling pathway - Towards a
novel concept of antiviral therapy. Biological Chemistry 389,
1307–1312 (2008).