1. How does influenza virus jump from animals to humans?
Haixu Tang
School of Informatics and Computing
Indiana University, Bloomington

2. Swine flu outbreak, 2009

3. Influenza pandemics in human history
Spanish flu, 1918
Most deadly natural disaster
Asian flu, 1957
Hong Kong flu, 1968
Seasonal flu: every year
Vaccine designed based on the observation of the flu strains spreading in animals

4. Influenza is cause by a virus
Orthomyxoviridae;
a class of RNA virus using RNA as genetic material;
globular article of a diameter ~100 nm;
Protected by a bilayer and matrix proteins;
~500 copies of H protein and ~100 copies of N proteins;
used for classification

5. Classification of influenza viruses
16 H genes and 9 genes found
Spanish flu, 1918: H1N1
Most deadly natural disaster
Asian flu, 1957: H2N2
Hong Kong flu, 1968: H3N2
Swine flu: 2009: H1N1
Current pandemic threat: H5N1
Several hundreds of active flu strains

6. Infection of flu viruses
Virus attached to the host cell;
Host cell
nucleus

7. Infection of flu viruses
Virus attached to the host cell;
Virus swallowed up by the host cell;

8. Infection of flu viruses
Virus is attached to the host cell;
Virus is swallowed up by the host cell;
Viral RNAs is released and enter the nucleus, where they are reproduced;

9. Infection of flu viruses
Virus is attached to the host cell;
Virus is swallowed up by the host cell;
Viral RNAs is released and enter the nucleus, where they are reproduced;
Fresh RNAs enter the cytosol;

10. Infection of flu viruses
Virus is attached to the host cell;
Virus is swallowed up by the host cell;
Viral RNAs is released and enter the nucleus, where they are reproduced;
Fresh RNAs enter the cytosol;
Viral RNAs act as mRNA to be translated into proteins forming new virus particles;

11. Infection of flu viruses
Virus is attached to the host cell;
Virus is swallowed up by the host cell;
Viral RNAs is released and enter the nucleus, where they are reproduced;
Fresh RNAs enter the cytosol;
Viral RNAs act as mRNA to be translated into proteins forming new virus particles;
New virus buds off from the membrane of the host cell;

12. Infection of flu viruses
Virus is attached to the host cell;
Virus is swallowed up by the host cell;
Viral RNAs is released and enter the nucleus, where they are reproduced;
Fresh RNAs enter the cytosol;
Viral RNAs act as mRNA to be translated into proteins forming new virus particles;
New virus buds off from the membrane of the host cell;

13. Spread of influenza viruses

14. Recognition of the virus to the host cell: hemagglutinin (H protein) vs. glycans
On the surface of animal cells, there exist a heavy coat of glycans (sugars linked to proteins or lipids);
Different animals may have glycans of different structures;
Human, pig and birds
Different influenza virus strains with different hemagglutinin proteins (a class of glycan binding protein) recognize glycans of different structures

15. Structure of glycans
linkage
monosaccharide

16. Some basic graph theory
Graph: modeling pairwise relation (edges) between subjects (nodes or vertices)
Tree: a graph with no cycle
Each node in a tree has zero (leaves) or more child nodes
Subtree: a subset of nodes/edge
Labels
Nodes: monosaccharides
Edges: linkage types

17. Animal glycans

18. Hemagglutinins recognize sialylated glycans
Human cells mainly express 2-6 linked sialylated glycans;
Bird cells mainly express 2-3 linked sialylated glycans;
Pig cells express both.
Vessel theory

19. Hemagglutinin-glycan interaction is more complicated
Several influenza strains were observed to be inconsistent with the theory
AV18 strain has hemagglutinin proteins recognize specifically 2-3 linked glycans, but are not transmissable in birds;
NY18 and Tx91 strains recognize both 2-3 and 2-6 linked glycans, but NY18 does not transmit efficiently in human population, whereas Tx91 does;
Some chimeric H1N1 strains with high binding affinity to 2-6 linked glycans, but do not spread efficiently in human and pig.

20. Experimental determination of binding specificities of hemagglutinins
Interaction assay
hemagglutinin
Glycan array
virus
The glycan motif finding problem

21. The glycan motif finding problem
Input: a set of (glycan) trees that are found to be recognized by a glycan binding protein (e.g. hemagglutinin)
Output: a l-treelet that is over-represented in the input set
l-treelet: a tree of a fixed small size l (e.g. l=4).
Over-representation: number of trees in the input set containing the treelet is much higher than the expected number in a random set of trees

22. Exhaustive counting of treelets

23. Finding over-represented treelets
Many 4-treelets appear in ALL input glycan trees.
Glycans are not random!

24. The glycan motif finding problem A different formulation
Input: a positive set of (glycan) trees that are found to be recognized by a glycan binding protein (e.g. hemagglutinin) and a negative set of glycan trees that are found NOT to be recognized by the same protein
Output: a l-treelet that is over-represented in the input positive set than the negative set
Over-representation: number of trees in the positive set containing the treelet is much higher than the number of trees in the negative set

25. Contingency table

26. Fisher’s exact test: significance test of the over-represented treelets
M glycans printed on the array
N positive, M-N negative
ni+ positives contain the treelet
ni- negatives contain the treelet + -
ni+
N-ni+
ni-
M-N-ni-
P=0.15

27. Glycan patterns found to be recognized by hemagglutinin
R. Sasisekharan and colleagues show is the pattern recognized by human influenza hemagglutinin, but not recognized by avian influenza hemagglutinin;
2-6 linked sialylated glycans with long oligosaccharide branch (with multiple lactosamine repeats) are predominantly expressed in the human upper respratory epithelial cells
Indeed, the binding specificity of hemagglutinin to 2-6 linked sialylated glycans is not sufficient for the spread of the influenza viruses in human populations.
Chandrasekaran A, et al. Nat Biotechnol , 2008; 26:107–113.

28. Spread of influenza viruses