Topological aspects of the human protein interaction network

It is currently widely accepted that the understanding of complex cell functions depends on an integrated network theoretical approach and not on an isolated view of the different molecular agents. Aim of this thesis was the examination of topological properties that mirror known biological aspects by depicting the human protein network with methods from graph- and network theory. The presented network is a partial human interactome of 9222 proteins and 36324 interactions, consisting of single interactions reliably extracted from peer-reviewed scientific publications. In general, one can focus on intra- or intermodular characteristics, where a functional module is defined as "a discrete entity whose function is separable from those of other modules". It is found that the presented human network is also scale-free and hierarchically organised, as shown for yeast networks before. The interactome also exhibits proteins with high betweenness and low connectivity which are biologically analyzed and interpreted here as shuttling proteins between organelles (e.g. ER to Golgi, internal ER protein translocation, peroxisomal import, nuclear pores import/export) for the first time. As an optimisation for finding proteins that connect modules, a new method is developed here based on proteins located between highly clustered regions, rather than regarding highly connected regions. As a proof of principle, the Mediator complex is found in first place, the prime example for a connector complex. Focusing on intramodular aspects, the measurement of k-clique communities discriminates overlapping modules very well. Twenty of the largest identified modules are analysed in detail and annotated to known biological structures (e.g. proteasome, the NFκB-, TGF-β complex). Additionally, two large and highly interconnected modules for signal transducer and transcription factor proteins are revealed, separated by known shuttling proteins. These proteins yield also the highest number of redundant shortcuts (by calculating the skeleton), exhibit the highest numbers of interactions and might constitute highly interconnected but spatially separated rich-clubs either for signal transduction or for transcription factors. This design principle allows manifold regulatory events for signal transduction and enables a high diversity of transcription events in the nucleus by a limited set of proteins. Altogether, biological aspects are mirrored by pure topological features, leading to a new view and to new methods that assist the annotation of proteins to biological functions, structures and subcellular localisations. As the human protein network is one of the most complex networks at all, these results will be fruitful for other fields of network theory and will help understanding complex network functions in general.

Effects of low-density lipoprotein receptor-related protein 4 and apolipoprotein E isoforms on bone metabolism in vivo

LRP4, member of the LDLR family, is a multifunctional membrane-bound receptor that is expressed in various tissues. The expression of LRP4 by osteoblasts, its novel interaction with Wnt-signaling inhibitors Dkk1 and SOST, and the lower levels of activated beta-catenin in different bone locations described here, adds another player to the long list of established factors that modulate canonical Wnt-signaling in bone. By demonstrating that in addition to Wise, LRP4 is able to interact with two additional important modulators of Wnt- and BMP-signaling, our perspective of the complexity of the integration of BMP and Wnt-signaling pathways on the osteoblast surface has expanded further. Nevertheless the recently described association of both the SOST and LRP4 genes with BMD in humans, together with our findings suggest that LRP4 plays a physiologically important role in the skeletal development and bone metabolism not only in rodents, but in humans as well. The efficiency with which LRP4 binds both SOST and Dkk1, presumably at the osteoblastic surface, LRP4 may act as a sink and competes with LRP5/6 for the binding of these Wnt antagonists, which then are no longer available for suppression of the signal through the LRP5/6 axis. rnApoE, a 299 amino acid glycoprotein, is a crucial regulator in the uptake of triglyceride, phospholipids, cholesteryl esters, and cholesterol into cells. ApoE has been linked to osteoporosis, and such a role is further strengthened by the present of a high bone mass phenotype in ApoE null mice. Until recently, the effects of respective ApoE isoforms E2, E3, and E4, and their impact on bone metabolism, have been unclear. Here we report that respective human ApoE knockin mice display diverse effects on bone metabolism. ApoE2 mice show decreased trabecular bone volume per total volume in femoral bone and lumbar spine in comparison to ApoE3 and E4 animals. In this context, urinary bone resorption marker DPD is increased in these animals, which is accompanied by a low ratio of osteoclastogenesis markers OPG/RANKL. Interestingly, serum bone formation markers ALP and OCN are diminished in ApoE4 mice. In contrast to this finding, ApoE2 mice show the lowest bone formation of all groups in vivo. These findings cannot be explained by the low receptor-affinity of ApoE2 and subsequent decreased uptake of triglyceride-rich lipoproteins by osteoblasts, resulting in elevated levels of undercarboxylated osteocalcin. Thus, other crucial pathways relevant for bone metabolism, e. g. Wnt/beta-catenin-signaling pathways, must be, compared to the ApoE3/4 isoforms, more affected by the ApoE2 isoform.