New technologies allow a census of an organism’s proteins and their interactions Recent advances in high-throughput technologies have allowed unprecedented insights into the entire protein content of cells and the interactions among proteins. The library of all proteins encoded in the genome of a cell is termed the proteome. Many proteins are presently recognized only on the basis of portions of sequenced genomes; they have not yet been isolated as biochemical entities, and practically nothing is known about their possible functions. A huge amount of effort now goes into using available sequence information, at the gene and protein levels, and structural data to predict functions of proteins with closely related sequences. These predictions then have to be verified by actual biochemical experimentation.
Another significant challenge is to classify the proteome of a given species into different protein classes. Huge databases are created by the international scientific community to create orderly classifications and derive biologically relevant insights into these seemingly chaotic data. Figure 1 illustrates the classification of the entire human proteome into different classes related to the subcellular localization of the proteins, their molecular function, and the biological processes in which the proteins participate. The known proteomes of some other model species, such as the yeast Saccharomyces cerevisiae, the worm Caenorhabditis elegans, and the fruit fly Drosophila melanogaster, have also been classified into these same categories. It is important to realize that these achievements, as impressive as they are, constitute only the first steps in understanding the way the proteome functions.
Another state-of-the-art development in the proteome field is the creation of interactome maps. These are databases that identify, through high-throughput experimental methods, all of the possible binary interactions, that is, interactions between any two proteins, in the entire proteome. Other databases identify inter actions in protein complexes, involving many protein partners. Some interactome maps focus on particular proteins or pathways—Figure 1 shows an example— while others are more ambitious, attempting to cover large portions of proteomes. Figure2 presents a recent interactome of a large portion of the yeast proteome; Figure 3 shows different representations of a portion of the human interactome. Clearly, the range of interactions is wonderfully complex, even with our presently limited knowledge.
Fig1. Partial interaction network for the protein Nanog. Nanog is known to participate in defining the pluripotent state of embryonic stem (ES) cells, that is, their ability to differentiate into numerous different cell types, depending on the applied stimuli. Affinity-purified Nanog partners were identified by mass spectrometry; in an iterative fashion, protein partners of these primary partners were then identified. The new protein partners were further verified experimentally by alternative techniques such as co-immunoprecipitation. The network is highly enriched in nuclear factors that are individually critical for the pluripotent state of the ES cell; the presence and levels of these factors are coordinated during further differentiation of the stem cells into different cell types. The tight protein network seems to function as a cellular module dedicated to pluripotency. The lines of different color that connect the proteins in the network indicate the methods by which the interaction has been verified experimentally. The circles of different colors denote proteins whose presence is essential for a particular stage of development, with the white circles designating proteins whose absence has not been correlated with developmental defects. [From Wang J, Rao S, Chu J et al. (2006) Nature 444: 364–368. With permission from Macmillan Publishers, Ltd.]
Fig2. Yeast protein interaction maps representing binary protein–protein interactions. Binary protein–protein interactions are those between two proteins. These maps are derived from two different high throughput experimental methods. Individual proteins are represented by dots, and interactions are shown by lines connecting the dots. (A) This interaction map from October 2008 combines three high-quality proteome-wide data sets, obtained by yeast two-hybrid screening. The map covers ~20% of the entire binary yeast interactome and contains 2,930 interactions among 2,018 proteins. (B) Map of binary interaction extracted from co-complex interactome maps, which are obtained by high throughput co-affinity purification followed by mass spectrometry, a principle similar to that presented in Box 3.9. The map is composed of 9,070 associations between 1,622 proteins. The two maps constructed by use of different system biology tools are fundamentally different but complementary in nature: they reflect different connections and biological properties of the interacting partners. Certain proteins are highly connected, with numerous partners; these are termed hub proteins. The absence of such proteins leads to a number of different phenotypes. Interestingly, hub proteins are characterized by intrinsic disorder, which may form the molecular basis for their promiscuous interaction with multiple partners. Interacting with different partners in different conditions constitutes the molecular basis for pleiotropic effects, that is, different roles of the same protein in multiple biochemical pathways. [From Yu H, Braun P, Yildirim MA et al. (2008) Science 322: 104–110. With permission from American Association for the Advancement of Science.]
Fig3. Different graphic representations of portions of the human interactome. These styles of representation are frequently used, but there are numerous ways to graph experimental interactome data. The estimated size of the binary interactome ranges from 130,000 to ~650,000 interactions; only ~8% of them have been identified. Ambiguity about the size of the human interactome stems from difficulties in identifying which of the possible biophysical interactions actually occur in cells. With the presently estimated number of human proteins, nearly 250 million protein pairs need to be tested to map the whole human interactome: such a huge undertaking is possible only with application of high-throughput approaches. The nematode Caenorhabditis elegans appears to have a protein number similar to that in humans, and the sizes of the two interactomes appear to be of the same order. Then, what makes humans different from worms? We do not have the answer yet. [A, courtesy of Seth Berger, Mount Sinai School of Medicine, New York, NY, USA. B, courtesy of Keiono and Andrew Garrow (creator of dataset), Wikimedia.]
Again, it must be noted that creating these beautiful-looking maps is only the first step in a process that tries to make sense of all this complexity. In fact, attempts to represent large, complex databases in a meaningful way, in both science and technology, have given rise to an entirely new discipline called visual analytics. Scientists not only seek ways to graphically represent an interactome (see Figure 2 and Figure 3) but also try to detect “what’s in there,” to extract the basic principles that regulate the interactions of individual components. The insights that interactome mapping is expected to provide into the functioning of living organisms cannot be overstated.
It is obvious that the enormous diversity of proteins that exist in living organ isms demands that every organism, in fact every cell, must have some repository of information to specify the primary structures of many proteins. The expression of this information must be under strict control as different cell types, even within a single organism, need different proteins to function, and these needs will change with growth and development. There must also be a way to preserve and transmit this information from generation to generation. That is the focus of this book.
