ANIMAL BIODIVERSITY AND EMERGING DISEASES
“Pathogeno-Proteomics” Toward a New Approach of Host–Vector–Pathogen Interactions Philippe Holzmuller,a Pascal Gr´ebaut,a Jean-Paul Brizard,b David Berthier,a Isabelle Chantal,a G´eraldine Bossard,a Bruno Bucheton,a Frederic Vezilier,a Paul Chuchana,a Rachel Bras-Gonc¸alves,a Jean-Loup Lemesre,a Philippe Vincendeau,c G´erard Cuny,a Roger Frutos,a and David G. Birond a b c
CIRAD UMR 17 [UMR 177 IRD-CIRAD], Montpellier, France
UMR5096 CNRS-IRD-Universit´e de Perpignan, Montpellier, France
Laboratoire de Parasitologie (EA 3677), Universit´e Victor Segalen, Bordeaux, France d
PIAF, UMR 547 INRA-Universit´e Blaise Pascal, Clermont-Ferrand, France
Many scientists working on pathogens (viruses, bacteria, fungi, parasites) are betting heavily on data generated by longitudinal genomic–transcriptomic–proteomic studies to explain biochemical host–vector–pathogen interactions and thus to contribute to disease control. Availability of genome sequences of various organisms, from viruses to complex metazoans, led to the discovery of the functions of the genes themselves. The postgenomic era stimulated the development of proteomic and bioinformatics tools to identify the locations, functions, and interactions of the gene products in tissues and/or cells of living organisms. Because of the diversity of available methods and the level of integration they promote, proteomics tools are potentially able to resolve interesting issues specific not only to host–vector–pathogen interactions in cell immunobiology, but also to ecology and evolution, population biology, and adaptive processes. These new analytical tools, as all new tools, contain pitfalls directly related to experimental design, statistical treatment, and protein identification. Nevertheless, they offer the potency of building large protein–protein interaction networks for in silico analysis of novel biological entities named “interactomes,” a way of modeling host–vector–pathogen interactions to define new interference strategies. Key words: host–vector–pathogen interactions; proteomics; interactome; modeling
Introduction Relationships between pathogens and their hosts and vectors depend on a molecular dialogue tightly regulated. Variability and crossregulation increase from genomic DNA (mutations, rearrangement, methylations) through Address for correspondence: Philippe Holzmuller, CIRAD UMR 17 Trypanosomes [UMR IRD-CIRAD 177 Interactions Hˆotes-VecteursParasites dans les trypanosomoses], TA A-17/G, Campus International de Baillarguet, 34398 Montpellier cedex 5. Voice: +33 (0) 4 67 59 37 49; fax: +33 (0) 4 67 59 38 94.
[email protected]
RNA transcripts (initiation, splicing, maturation, editing, stability) to functional proteins (initiation, folding, post-translational modifications, localization, function). The analytical levels of host–vector–pathogen interactions by themselves exhibit cross-talk from the constitutive to the expressive and finally to the functional level (Fig. 1A). During host–vector–pathogen interactions, all analytical levels (genome, transcriptome, proteome: whole cell content, and secretome: naturally excreted–secreted molecules) interact with
Animal Biodiversity and Emerging Diseases: Ann. N.Y. Acad. Sci. 1149: 66–70 (2008). C 2008 New York Academy of Sciences. doi: 10.1196/annals.1428.061
66
Holzmuller et al.: Pathogeno-Proteomics
Figure 1. Cross-talk between the analytical levels within an organism (A) and during host–vector– pathogen interactions (B).
each other within one actor, then between two, and finally, in the case of vectorial transmission, a third level of interactions complete the dialogues at all analytical levels between hosts, vectors, and pathogens (Fig. 1B). We propose a new integrative approach, “pathogenoproteomics,” to study the cross-talk in the host– vector–pathogen associations. We review here the concept, the key steps, and the advantages and pitfalls of bioinformatics tools to demonstrate essential gene products involved in the molecular dialogue occurring during the host– vector–pathogen interactions.
Number of Experimental Treatments and Design and Preparation of Biological Material The design and preparation of the biological material you wish to study is directly correlated to the question you wish to answer. The experimental treatments must include all pos-
67
Figure 2. Key steps involved in “pathogenoproteomics” based on the new integrative approach.
sible variations not due to the biological process you want to pinpoint. For example, two treatments could be enough to compare two strains of pathogen, whereas at least nine treatments would be necessary to study host response to one pathogen strain, each treatment including replicates. Once the number of experimental treatments is defined, the second step consists in designing the biological samples corresponding to the treatments. For example in trypanosomosis, to consider the variability in host and parasite responses during infection, a strong control is needed with two kinds of interactions: one making the host species differ and the second making the parasite species differ (Fig. 2). This generates replicative samples for proteomic analysis. Considering queries on molecular cross-talk during infection, proteomics tools will allow identification of differentially expressed proteins in both host and parasite proteomes. Their molecular characterization will serve to categorize them
68
Annals of the New York Academy of Sciences
Figure 3. Comparison of proteomic tools.
according to specific or nonspecific, and constitutive or inducible properties (Fig. 2). Proteomics Tools and Their Pitfalls The era of proteomics allows us to study directly the molecular dialogue occurring during host–vector–pathogen interactions. Methods of reference are two-dimensional electrophoresis (2-DE) separating molecules, followed by mass spectrometry (MS) identification. Other techniques based on liquid chromatography followed by MS can give greater resolution, especially for low-molecular-weight proteins. Some binding techniques as protein arrays can also be developed for high throughput.1 Advantages and disadvantages of available techniques are summarized in Figure 3. Bioinformatics: The Good, the Bad and the Dodgy The “omics” (e.g., genomics, transcriptomics, and proteomics) tools generate an important amount of data. For example, a single
proteomics experiment reveals the expression information for hundred or thousands of proteins. Therefore, data analysis (bioinformatics) is an essential part of this type of research. In fact, bioinformatics analysis in many cases takes more time than the actual experiment and requires special skills and tools. Over the past decade, an important number of commercial softwares involving ever-more powerful algorithms and statistical tools than those of the previous generations have been designed to help researchers deal with the sheer quantity of data produced. Two-dimensional electrophoresis is still a powerful separation technique which allows simultaneous resolution of thousands of proteins contained in a proteome of an organism.1,2 All 2-DE software ensures fast and reliable gel comparison, and they are now capable of multiple gel analysis, including filtering of 2-DE images, automatic spot detection, normalization of the volume of each protein spot, and differential and statistical analysis.3–5 These helpful bioinformatics tools, which allow the differential expression of a given proteome (cell, tissue, or fluid of an organism) between different treatments and/or between
69
Holzmuller et al.: Pathogeno-Proteomics
populations to be revealed, aim to find and characterize proteins linked to particular biological phenomena. However, some studies reveal that human intervention is necessary to correct the step of detection of candidate protein spots (i.e., deletion of false protein spots and correction of the shape of protein spots) as well as the pairing step of protein spots within a same class (category) of gels and between different classes (categories) of gels. Many kinds of 2-DE software offer the Student t-test and one-way and two-way ANOVA as statistical tools to detect significant alteration in protein expression. However, an inappropriate utilization of these tests may result in an important number of false-positive findings, which is effectively the case in more than 60% of recent proteomics studies.1 Moreover, assuming a 5% level of error (i.e., P = 0.05) for 2000 protein spots means that there are potentially some 100 (i.e., 0.05 × 2000) false-positive results, which is unacceptable. Multiple testing correction methods, such as the Bonferroni correction and testing for the false discovery rate (FDR), adjust the Student t-test or ANOVA values for each protein spot to keep the overall error rate as low as possible. The Bonferroni correction multiplies the unadjusted P-values by the total number of tests performed. The FDR is a less stringent correction method, but a more practical approach than the Bonferonni correction. The FDR is defined as V/R for R > 0 and FDR = 0 if R = 0 (where V denotes the number of falsely rejected hypotheses and R indicates the total number of rejected hypotheses). Since V is unobserved, a sequential P-values procedure has been developed to control the expected value of the FDR [i.e., E(FDR)] under the assumption that the test statistics are independent. The resulting process controls E(FDR) at the fixed level α for any joint distribution of the P-values. Actually, considering the amount of data generated, the integrative approaches afforded by “omics” studies demand the involvement of bioinformatics researchers to warranty the analytical interpretation of biological phenomena.
The Interactome: Toward Modelization of Host–Vector–Pathogen Interactions The last few years have witnessed the birth of new biological entities named “interactomes.” They correspond in an “ideal world” to the complete set of protein–protein interactions existing between all the proteins of an organism. In reality they are far to from complete since an unknown number of interactions are yet to be discovered. Current interactomes are only a part of the whole set of possible interactions occurring within and between an organism or organisms. Although the deciphering of the interactomes of the main model organisms is not yet complete, studies of the interactomes of pathogens are increasing. The first pathogens to be investigated in terms of their interactomes were the hepatitis C virus6 and Helicobacter pylori.7 More recently still, the interactomes of the malaria parasite, Plasmodium falciparum, have been determined.8 This makes one believe that in the near future, as initiated by Uetz and colleagues,9 the docking of the interactomes of pathogens onto those of their hosts will soon be possible. In the near future, a greater amount of proteomics data will be available for many organisms and will in turn open up new prospects for interactome studies. By way of example, recently the combination of the proteomics and interactome data on the human nuclear proteome permitted function to be assigned to 49 previously uncharacterized human nucleolar proteins and to reveal the first draft of the human ribosome biogenesis pathway.10 Concluding Remarks Despite the pitfalls they include, 2-DE and other proteomic tools are the functional steps of integrative gen-transcipt-prote-omics approaches. They open a promising research field with the study of the interactome of organisms along with the instantaneous and the
70
Annals of the New York Academy of Sciences
temporal interactomes resulting from the interaction of the proteomes between organisms, especially, as far as we are concerned, host–vector–pathogen interactions. The analysis of “docked interactomes” is certainly a very promising and exciting aspect of interactomics because of its obvious potential impact on human and animal health. Acknowledgments
This work was supported by grants from the European Community (INCO-DEV Trypadvac 2 PL003716) and from CIRAD and IRD. Conflicts of Interest
The authors declare no conflicts of interest. References 1. Biron, D.G. et al. 2006. The pitfalls of proteomics experiments without the correct use of bioinformatics tools. Proteomics 6: 5577–5596.
2. Rabilloud, T. 2002. Two-dimensional gel electrophoresis in proteomics: old, old fashioned, but it still climbs up the mountains. Proteomics 2: 3– 10. 3. Barrett, J. et al. 2005. Analysing proteomic data. Int. J. Parasitol. 35: 543–553. 4. Marengo, E. et al. 2005. Numerical approaches for quantitative analysis of two-dimensional maps: a review of commercial software and home-made systems. Proteomics 5: 654–666. 5. Wheelock, A.M. & S. Goto. 2006. Effects of postelectrophoretic analysis on variance in gel-based proteomics. Expert. Rev. Proteomics 3: 129–142. 6. Flajolet, M. et al. 2000. A genomic approach of the hepatitis C virus generates a protein interaction map. Gene 242: 369–379. 7. Rain, J.C. et al. 2001. The protein-protein interaction map of Helicobacter pylori. Nature 409: 211– 215. 8. LaCount, D.J. et al. 2005. A protein interaction network of the malaria parasite Plasmodium falciparum. Nature 438: 103–107. 9. Uetz, P. et al. 2006. Herpesviral protein networks and their interaction with the human proteome. Science 311: 239–242. 10. Hinsby, A.M. et al. 2006. A wiring of the human nucleolus. Mol. Cell 22: 285–295.
