NETWORK TOPOLOGY IN HUMAN PROTEIN INTERACTION DATA PREDICTS FUNCTIONAL ASSOCIATION

High-throughput assays, such as yeast two-hybrid system, have generated a huge amount of protein-protein interaction (PPI) data in the past decade. This tremendously increases the need for developing reliable methods to systematically and automatically suggest protein functions and relationships between them. With the available PPI data, it is now possible to study the functions and relationships in the context of a large-scale network. To data, several network-based schemes have been provided to effectively annotate protein functions on a large scale. However, due to those inherent noises in high-throughput data generation, new methods and algorithms should be developed to increase the reliability of functional annotations. Previous work in a yeast PPI network (Samanta and Liang, 2003) has shown that the local connection topology, particularly for two proteins sharing an unusually large number of neighbors, can predict functional associations between proteins, and hence suggest their functions. One advantage of the work is that their algorithm is not sensitive to noises (false positives) in high-throughput PPI data. In this study, we improved their prediction scheme by developing a new algorithm and new methods which we applied on a human PPI network to make a genome-wide functional inference. We used the new algorithm to measure and reduce the influence of hub proteins on detecting functionally associated proteins. We used the annotations of the Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) as independent and unbiased benchmarks to evaluate our algorithms and methods within the human PPI network. We showed that, compared with the previous work from Samanta and Liang, our algorithm and methods developed in this study improved the overall quality of functional inferences for human proteins. By applying the algorithms to the human PPI network, we obtained 4,233 significant functional associations among 1,754 proteins. Further comparisons of their KEGG and GO annotations allowed us to assign 466 KEGG pathway annotations to 274 proteins and 123 GO annotations to 114 proteins with estimated false discovery rates of <21% for KEGG and <30% for GO. We clustered 1,729 proteins by their functional associations and made pathway analysis to identify several subclusters that are highly enriched in certain signaling pathways. Particularly, we performed a detailed analysis on a subcluster enriched in the transforming growth factor β signaling pathway (P<10-50) which is important in cell proliferation and tumorigenesis. Analysis of another four subclusters also suggested potential new players in six signaling pathways worthy of further experimental investigations. Our study gives clear insight into the common neighbor-based prediction scheme and provides a reliable method for large-scale functional annotations in this post-genomic era.

CART: an Hrs/actinin-4/BERP/myosin V protein complex required for efficient receptor recycling.

Altering the number of surface receptors can rapidly modulate cellular responses to extracellular signals. Some receptors, like the transferrin receptor (TfR), are constitutively internalized and recycled to the plasma membrane. Other receptors, like the epidermal growth factor receptor (EGFR), are internalized after ligand binding and then ultimately degraded in the lysosome. Routing internalized receptors to different destinations suggests that distinct molecular mechanisms may direct their movement. Here, we report that the endosome-associated protein hrs is a subunit of a protein complex containing actinin-4, BERP, and myosin V that is necessary for efficient TfR recycling but not for EGFR degradation. The hrs/actinin-4/BERP/myosin V (CART [cytoskeleton-associated recycling or transport]) complex assembles in a linear manner and interrupting binding of any member to its neighbor produces an inhibition of transferrin recycling rate. Disrupting the CART complex results in shunting receptors to a slower recycling pathway that involves the recycling endosome. The novel CART complex may provide a molecular mechanism for the actin-dependence of rapid recycling of constitutively recycled plasma membrane receptors.

Cloning and characterization of the mouse ret finger protein (rfp), a B box zinc finger protein

An important question in biology is to understand the role of specific gene products in regulating embryogenesis and cellular differentiation. Many of the regulatory proteins possess specific motifs, such as the homeodomain, basic helix-loop-helix structure, zinc finger, and leucine zipper. These sequence motifs participate in specific protein-DNA, protein-RNA, and protein-protein interactions, and are important for the function of these regulatory proteins.^ The human rfp (ret finger protein) belongs to a novel zinc finger protein family, the B box zinc finger family. Most of the B box proteins, including rfp, have a conserved tripartite motif, consisting of two novel zinc fingers (the RING finger and the B box) and a coiled-coil domain. Interestingly, a fusion protein between the tripartite motif of rfp and the tyrosine kinase domain of c-ret has transforming activity. In this study, we examined the expression of rfp during mouse development, and characterized the role of the tripartite motif in rfp function.^ We cloned the mouse rfp cDNA, which shares a 98.4% homology with the human sequence at amino acid level. Such strikingly high degree of homology indicates the high evolutionary pressure on the conservation of the sequence, suggesting that rfp may have an important function. Using the somatic cell hybrid system, we assigned the rfp gene to mouse chromosome 13 and human chromosome 6. Rfp transcripts and protein were ubiquitous in day 10.5-13.5 mouse embryos; however, they were restricted in adult mice, with the highest level of expression in the testis. Rfp expression in the testis is detected only in late pachytene spermatocytes and round spermatids. In both embryos and spermatogenic cells, rfp protein was distributed within cell nuclei in a punctate pattern, similar to the PODs (PML oncogenic domains) observed with another B box protein, PML. In cultured mammalian cells, we found that rfp was indeed co-localized to the PODs with PML. Using the yeast two-hybrid system, we showed that the rfp could specifically interact with PML, and that the interaction was dependent on the distal portion of the rfp coiled-coil domain.^ We also showed that rfp could form homodimers, and both the B box and coiled-coil domain were required for proper dimerization. It seems that the proximal portion of the coiled-coil domain provides the interacting interface, while the B box zinc finger orients the coil and maintains the correct structure of the whole molecule. Our data are consistent with the zinc-binding property and structural analysis of the B box. The RING finger seems to be involved in rfp nuclear localization through interaction with other proteins. We believe that homodimerization and interaction with PML are important for the normal interaction of rfp during development and differentiation. In addition, rfp homodimerization may also be essential for the oncogenic activation of the rfp-ret fusion protein. ^

Biochemical characterization of PICK1, a PKC binding protein

Protein kinase C (PKC) is a family of serine-threonine kinases that are activated by a wide variety of hormones, neurotransmitters and growth factors. A single cell type contains multiple isoforms that are translocated to distinct and different subcellular sites upon mitogenic stimulus. Many different cellular responses are attributed to PKC activity though relatively few substrates or binding proteins have been definitively characterized. We used the hinge and catalytic domain of PKC$\alpha$ (PKC7) in a yeast two-hybrid screen to clone proteins that interact with C-kinase (PICKs). One protein which we have termed PICK1 may be involved in PKC$\alpha$-specific function at the level of the nuclear membrane after activation. Binding of PICK1 to PKC$\alpha$ has been shown to be isoform specific as it does not bind to PKC$\beta$II or PKC$\alpha$ in the yeast two-hybrid system. PICK1 mRNA expression level is highest in testis and brain with lower levels of expression in skeletal muscle, heart, kidney, lung and liver. PICK1 protein contains five PKC consensus phosphorylation sites and serves as an in vitro substrate for PKC. The PICK1 protein also contains a P-Loop motif that has been shown to bind ATP or GTP in the Ras family of oncoproteins as well as the G-Protein family. Proteins which bind ATP or GTP using this motif all have some sort of catalytic function although none has been identified for PICK1 as yet. PICK1 contains a DHR/GLGF motif at the N-terminus of the protein. The DHR/GLGF motif is contained in a number of recently described proteins and has been shown to mediate protein-protein interactions at the level of membranes and cytoskeleton. When both PKC$\alpha$ and PICK1 are co-expressed in Cos1 cells the two proteins co-localize to the perinucleus in immunoflouresence studies and co-immunoprecipitate. The binding site for PKC7 has been localized to amino acids 1-358 on PICK1 which contains the DHR/GLGF motif. Binding of PICK1 to PKC$\alpha$ requires the hinge and C-terminal domains of PKC$\alpha$. In vitro, PICK1 binds to PKC$\alpha$ and inhibits its activity as assayed by myelin basic protein phosphorylation. PICK1 also binds to TIS21, a primary response gene that is expressed in response to phorbol ester and growth factor treatment. The Caenorhabditis elegans homologue of PICK1 has been cloned and sequenced revealing a high degree of conservation in the DHR/GLGF motif. A more C-terminal region also shows a high degree of conservation, and the C. elegans PICK1 homologue binds to PKC7 suggesting a conservation of function. Taken together these results suggest that PICK1 may be involved in a PKC$\alpha$-specific function at the level of the nuclear membrane. ^

Cloning and genetic characterization of Skb1: A novel regulator of Shk1 and mitosis in Schizosaccharomyces pombe

A partial skb1 gene was originally isolated in a yeast two-hybrid screen for Shk1-interacting polypeptides. Shk1 is one of two Schizosaccharomyces pombe p21Cdc42/Rac-activated kinases (PAKs) and is an essential component of the Ras1-dependent signal transduction pathways regulating cell morphology and mating responses in fission yeast. After cloning the skb1 gene we found the Skb1 gene product to be a novel, nonessential protein lacking homology to previously characterized proteins. However the identification of Skb1 homologs in C. elegans, S. cerevisiae, and H. sapiens reveals evolution has conserved the skb1 gene. Fission yeast cells carrying a deletion of skb1 exhibit a defect in cell size but not mating abilities. This defect is suppressed by high copy shk1. Fission yeast overexpressing skb1 were found to undergo cell division at a length 1.5X greater than normal. In the two-hybrid system, Skb1 interacts with a subdomain of the Shk1 regulatory region distinct from that with which Cdc42 interacts, and forms a ternary complex with Shk1 and Cdc42. By use of yeast genetics, we have established a role for Skb1 as a positive regulator of Shk1. Co-overexpression of shk1 with skb1 was found to suppress the morphology defect, but not the sterility, of ras1Δ fission yeast. Thus, the function of Skb1 is restricted to a morphology control pathway. We determined that Skb1 functions as a negative regulator of mitosis and does this through a Shk1-dependent mechanism. The mitotic regulatory function of Skb1 and Shk1 was also partially dependent upon Wee1, a direct negative regulator of the cyclin-dependent kinase Cdc2. The role for Skb1 and Shk1 as mitotic regulators is the first connection from a PAK protein to control of the cell cycle. Furthermore, Skb1 is the first non-Cdc42/Rac PAK modulator to be identified. ^

Characterization of hLodestar: A CDC5L interacting protein

Lodestar, a Drosophila maternal-effect gene, is essential for proper chromosome segregation during embryonic mitosis. Mutations in lodestar cause chromatin bridging in anaphase, preventing the sister chromatids from fully separating and leaving chromatin tangled at the metaphase plate. Drosophila lodestar protein was originally identified, in purified fractions of Drosophila Kc cell nuclear extracts, by its ability to suppress the generation of long RNA polymerase II transcripts. The human homolog of this protein (hLodestar) was cloned and studied in comparison to the Drosophila lodestar activities. The results of these studies show, similar to the Drosophila protein, hLodestar has dsDNA-dependent ATPase and transcription termination activity in vitro. hLodestar has also been shown to release RNA polymerase I and II stalled at a cyclobutane thymine dimer. Lodestar belongs to the SNF2 family of proteins, which are members of the DExH/D helicase super-family. The SNF2 family of proteins are believed to play a critical role in altering protein-DNA interactions in a variety of cellular contexts. We have recently isolated a human cDNA (hLodestar) that shares significant homology to the Drosophila lodestar gene. The 4.6 kb clone contains an open reading frame of 1162 amino acids, and shares 55% similarity and 46% identity to the Drosophila Lodestar protein sequence. Our studies looking for hLodestar interacting proteins revealed an association with CDC5L in the yeast two-hybrid system and co-immunoprecipitation experiments. CDC5L has been well documented to be a component of the spliceosome. Our data suggests hLodestar is involved in splicing through in vitro assembly and splicing reactions, in addition to its association with spliceosomes purified from HeLa nuclear extract. Although many other members of the DExH/D helicase super-family have been linked to splicing, this is the first SNF2 family member to be implicated in the splicing reaction. ^

CCTeta: A novel soluble guanylyl cyclase-interacting protein

Nitric oxide (NO) transduces most of its biological effects through activation of the heterodimeric enzyme, soluble guanylyl cyclase (sGC). Activation of sGC results in the production of 3′,5 ′-cyclic guanosine monophosphate (cGMP) from 5′ -guanosine triphosphate (GTP). In this thesis, we demonstrate a novel protein interaction between CCT (chaperonin containing t-complex polypeptide) subunit η and the α1β1 isoform of sGC. Using the yeast-two-hybrid system, CCTη was found to interact with the N-terminal portion of β1 subunit of sGC. This interaction was then confirmed in vitro with a co-immunoprecipitation from mouse brain. The interaction between these two proteins was further supported by a co-localization of the proteins within rat brain. Using the yeast-two-hybrid system, CCTη was found to bind to the N-terminal portion of sGC. In vitro assays with purified CCTη and Sf9 lysate expressing sGC resulted in a 33% inhibition of sodium nitroprusside (SNP)-stimulated sGC activity. The same assays were then performed using BAY41-2272, an NO-independent allosteric sGC activator, and CCTη had no effect on this activity. Furthermore, CCTη had no effect on the activity of αβCys105 sGC a constitutively active mutant that lacks a heme group. Of note is the fact that the full-length CCTη-expressing bacterial lysate inhibited the activity of sGC-expressing Sf9 lysate by 48% compared with GST alone. This indicates that the amino terminal 94 amino acids of CCTη are important to the inhibition of sGC activity. Lastly, a 45% inhibition of sGC activity by CCTη was seen in vivo in BE2 cells stably transfected with CCTη and treated with SNP. The fact that the inhibition of sGC was more pronounced with bacterial lysate expressing CCTη versus the purified CCTη implies that some factor in the bacterial lysate enhances the inhibitory effect of CCTη. Because the level of inhibition seen in bacterial lysate and in vivo experiments is similar, might imply that the factor that aids in CCTη effect on sGC is conserved. Together, these data suggest that CCTη is a novel type of sGC inhibitor that inhibits sGC by modifying the binding of NO to the heme group or the subsequent conformational changes induced by NO binding. ^

Identification and characterization of Xenopus Kazrin, a nucleocytoplasmic shuttling protein that interacts with ARVCF catenin

In common with other members of the p120-catenin subclass of catenins, ARVCF-catenin appears to have multiple cellular and developmental functions. In Xenopus, our lab recently demonstrated that xARVCF- and Xp120-catenins are each essential for early vertebrate embryogenesis, being functionally linked to Rho-family GTPases (RhoA, Rac) and cadherin metabolic stability. For the project described here, the yeast two-hybrid system was employed to screen a Xenopus laevis neurula library for proteins that interact with xARVCF, resulting in the identification of the Xenopus homolog of Kazrin (xKazrin). Kazrin is a variably-spliced protein of unknown function that has been shown to interact with periplakin and envoplakin, components of desmosomal junctions. Kazrin's primary sequence is highly conserved across vertebrate species and is composed of an amino-terminal nuclear export sequence (NES), a carboxy-terminal nuclear localization sequence (NLS) and a central predicted coiled-coil domain. In vitro and in vivo authenticity tests demonstrated that xARVCF-catenin interacts directly with xKazrin via xARVCF's Armadillo and carboxy-terminal regions and xKazrin's coiled-coil domain. The interaction of xARVCF-catenin with xKazrin is specific and does not extend to the related Xp120-catenin. xKazrin co-localized with E-cadherin at sites of cell-cell contact and could be co-immunoprecipitated with components of the cadherin complex. xKazrin was also present in the cytoplasm and nucleus. Suggestive of a nuclear role, mutation of xKazrin's predicted NLS resulted in nuclear exclusion, while deletion of the predicted NES resulted in loss of sensitivity to nuclear export inhibitors. Within Xenopus embryos, xKazrin was expressed across all developmental stages and appeared at varying levels in adult tissues. Morpholino depletion of xKazrin from Xenopus embryos resulted in axial elongation abnormalities and loss of tissue integrity after neurulation. Over-expression of xKazrin had no effect, while over-expression of a NLS mutant resulted in a mild phenotype similar to that seen in xKazrin depleted embryos. Interestingly, the axial phenotype resulting from reduced xKazrin levels was largely rescuable by xARVCF over-expression. In conjunction with xARVCF-catenin, xKazrin has properties consistent with its function at cell-cell contact sites and in the nucleus. ^

THE NEW ROLE OF PROPROTEIN CONVERTASE SUBTILISIN/KEXIN TYPE 9: A CONNECTION OF PROPROTEIN CONVERTASE SUBTILISIN/KEXIN TYPE 9, APOLIPOPROTEIN B, and AUTOPHAGY

Plasma low-density lipoprotein (LDL) levels are positively correlated with the incidence of coronary artery disease. In the circulation, the plasma LDL clearance is mainly achieved by the uptake via LDL receptor (LDLR). Proprotein convertase subtilisin/kexin type 9 (PCSK9) is a newly discovered gene, playing an important role in LDL metabolism. Gain-of-function mutations of PCSK9 lead to hypercholesterolemia and loss-of-function mutations of PCSK9 are associated with decrease of LDL cholesterol. The effects of PCSK9 on cholesterol levels are the consequence of a strong interaction between the catalytic domain of PCSK9 and epidermal growth factor-like repeat A (EGF-A) domain of LDLR on the cell surface of hepatocytes. This PCSK9/LDLR complex enters the cell via endocytosis, where both PCSK9 and LDLR are removed via the lysosome pathway, resulting in decreased levels of LDLR and accumulation of LDL in the plasma. However, whether this is the exclusive function of PCSK9 on LDL metabolism was challenged by us; we observed PCSK9 interacted with apolipoprotein B (apoB) and increased apoB production, irrespective of the LDLR. ApoB is the primary structure protein of LDL particle and it also serves as the ligand for the LDL receptor. There is ample evidence showing that the levels of apoB are a better indicator for heart disease than either total cholesterol or LDL cholesterol levels. We used a second-generation adenoviral vector to overexpress PCSK9 (Ad-PCSK9) in wild-type C57BL/6 and LDLR deficient mice (Ldlr-/- and Ldlr-/-Apobec1-/-). Our study revealed that overexpression of PCSK9 promoted the production and secretion of apoB in the form of very-low density lipoprotein (VLDL), which is the precursor of LDL, in the 3 mouse models studied (C57BL/6J, Ldlr-/-, and Ldlr-/-Apobec1-/-). The increased apoB production in mice was regulated at post-transcriptional levels, since there was no difference in apoB mRNA levels between mice treated with Ad-PCSK9 and control vector Ad-Null. By using pulse-chase experiment on primary hepatocytes, we showed that overexpression of PCSK9 increased the secretion of apoB, independent of LDLR. In the circulation, we showed that PCSK9 was associated with LDL particles. By using 3 different protein–protein interaction assays of co-immunoprecipitation, mammalian two-hybrid system, and in situ proximity ligation assay, we demonstrated a direct protein–protein interaction between PCSK9 and apoB. The impact of this interaction inhibited the physiological removal process of apoB via autophagosome/lysosome pathway in an LDLR-independent fashion, resulting in increased production and secretion of apoB-containing lipoproteins. The significance of this process was shown in the Pcsk9 knockout mice in the background of Ldlr-/-Apobec1-/- mice (triple knockout mice); in the absence of Pcsk9 (triple knockout mice) the levels of cholesterol, triacylglycerol, and apoB decreased significantly in comparison to that of Ldlr-/-Apobec1-/- mice. Taken together, our study demonstrated a direct intracellular interaction of PCSK9 with apoB, resulting in the inhibition of apoB degradation via the autophagosome/lysosome pathway independent of LDLR. This discovery provides a new concept of the importance of PCSK9 and suggests new approaches for the therapeutic intervention of hyperlipidemia.

Structural determinants of subunit -subunit interactions and subcellular localization of the calcium /calmodulin -dependent protein kinase II

The multifunctional Ca$\sp{2+}$/calmodulin-dependent protein kinase II (CaM kinase) is a Ser/Thr directed protein kinase that participates in diverse Ca$\sp{2+}$ signaling pathways in neurons. The function of CaM kinase depends upon the ability of subunits to form oligomers and to interact with other proteins. Oligomerization is required for autophosphorylation which produces significant functional changes that include Ca$\sp{2+}$/calmodulin-independent activity and calmodulin trapping. Associations with other proteins localize CaM kinase to specific substrates and effectors which serves to optimize the efficiency and speed of signal transduction. In this thesis, we investigate the interactions that underlie the appropriate positioning of CaM kinase activity in cells. We demonstrate that the subcellular distribution of CaM kinase is dynamic in hippocampal slices exposed to anoxic/aglycemic insults and to high K$\sp{+}$-induced depolarization. We determine the localization of CaM kinase domains expressed in neurons and PC-12 cells and find that the C-terminal domain of the $\alpha$ subunit is necessary for localization to dendrites. Moreover, monomeric forms of the enzyme gain access to the nucleus. Attempts made to identify novel CaM kinase binding proteins using the yeast two-hybrid system resulted in the isolation of hundreds of positive clones. Those that have been sequenced are identical to CaM kinase isoforms. Finally, we report the discovery of specific regions within the C-terminal domain that are necessary and sufficient for subunit-subunit interactions. Differences between the $\alpha$ and $\beta$ isoforms were discovered that indicate unique structural requirements for oligomerization. A model for how CaM kinase subunits interact to form holoenzymes and how structural heterogeneity might influence CaM kinase function is presented. ^

Evaluating the impact of vertical integration on health system peformance

Vertical integration is grounded in economic theory as a corporate strategy for reducing cost and enhancing efficiency. There were three purposes for this dissertation. The first was to describe and understand vertical integration theory. The review of the economic theory established vertical integration as a corporate cost reduction strategy in response to environmental, structural and performance dimensions of the market. The second purpose was to examine vertical integration in the context of the health care industry, which has greater complexity, higher instability, and more unstable demand than other industries, although many of the same dimensions of the market supported a vertical integration strategy. Evidence on the performance of health systems after integration revealed mixed results. Because the market continues to be turbulent, hybrid non-owned integration in the form of alliances have increased to over 40% of urban hospitals. The third purpose of the study was to examine the application of vertical integration in health care and evaluate the effects. The case studied was an alliance formed between a community hospital and a tertiary medical center to facilitate vertical integration of oncology services while maintaining effectiveness and preserving access. The economic benefits for 1934 patients were evaluated in the delivery system before and after integration with a more detailed economic analysis of breast, lung, colon/rectal, and non-malignant cases. A regression analysis confirmed the relationship between the independent variables of age, sex, location of services, race, stage of disease, and diagnosis, and the dependent variable, cost. The results of the basic regression model, as well as the regression with first-order interaction terms, were statistically significant. The study shows that vertical integration at an intermediate health care system level has economic benefits. If the pre-integration oncology group had been treated in the post-integration model, the expected cost savings from integration would be 31.5%. Quality indicators used were access to health care services and research treatment protocols, and access was preserved in the integrated model. Using survival as a direct quality outcome measure, the survival of lung cancer patients was statistically the same before and after integration. ^