Computational Inference of Genome-Wide Protein-DNA Interactions Using High-Throughput Genomic Data

Transcriptional regulation has been studied intensively in recent decades. One important aspect of this regulation is the interaction between regulatory proteins, such as transcription factors (TF) and nucleosomes, and the genome. Different high-throughput techniques have been invented to map these interactions genome-wide, including ChIP-based methods (ChIP-chip, ChIP-seq, etc.), nuclease digestion methods (DNase-seq, MNase-seq, etc.), and others. However, a single experimental technique often only provides partial and noisy information about the whole picture of protein-DNA interactions. Therefore, the overarching goal of this dissertation is to provide computational developments for jointly modeling different experimental datasets to achieve a holistic inference on the protein-DNA interaction landscape.

We first present a computational framework that can incorporate the protein binding information in MNase-seq data into a thermodynamic model of protein-DNA interaction. We use a correlation-based objective function to model the MNase-seq data and a Markov chain Monte Carlo method to maximize the function. Our results show that the inferred protein-DNA interaction landscape is concordant with the MNase-seq data and provides a mechanistic explanation for the experimentally collected MNase-seq fragments. Our framework is flexible and can easily incorporate other data sources. To demonstrate this flexibility, we use prior distributions to integrate experimentally measured protein concentrations.

We also study the ability of DNase-seq data to position nucleosomes. Traditionally, DNase-seq has only been widely used to identify DNase hypersensitive sites, which tend to be open chromatin regulatory regions devoid of nucleosomes. We reveal for the first time that DNase-seq datasets also contain substantial information about nucleosome translational positioning, and that existing DNase-seq data can be used to infer nucleosome positions with high accuracy. We develop a Bayes-factor-based nucleosome scoring method to position nucleosomes using DNase-seq data. Our approach utilizes several effective strategies to extract nucleosome positioning signals from the noisy DNase-seq data, including jointly modeling data points across the nucleosome body and explicitly modeling the quadratic and oscillatory DNase I digestion pattern on nucleosomes. We show that our DNase-seq-based nucleosome map is highly consistent with previous high-resolution maps. We also show that the oscillatory DNase I digestion pattern is useful in revealing the nucleosome rotational context around TF binding sites.

Finally, we present a state-space model (SSM) for jointly modeling different kinds of genomic data to provide an accurate view of the protein-DNA interaction landscape. We also provide an efficient expectation-maximization algorithm to learn model parameters from data. We first show in simulation studies that the SSM can effectively recover underlying true protein binding configurations. We then apply the SSM to model real genomic data (both DNase-seq and MNase-seq data). Through incrementally increasing the types of genomic data in the SSM, we show that different data types can contribute complementary information for the inference of protein binding landscape and that the most accurate inference comes from modeling all available datasets.

This dissertation provides a foundation for future research by taking a step toward the genome-wide inference of protein-DNA interaction landscape through data integration.

Mapping protein-DNA interactions using ChIP-sequencing

Chromatin immunoprecipitation (ChIP) allows enrichment of genomic regions which are associated with specific transcription factors, histone modifications, and indeed any other epitopes which are present on chromatin. The original ChIP methods used site-specific PCR and Southern blotting to confirm which regions of the genome were enriched, on a candidate basis. The combination of ChIP with genomic tiling arrays (ChIP-chip) allowed a more unbiased approach to map ChIP-enriched sites. However, limitations of microarray probe design and probe number have a detrimental impact on the coverage, resolution, sensitivity, and cost of whole-genome tiling microarray sets for higher eukaryotes with large genomes. The combination of ChIP with high-throughput sequencing technology has allowed more comprehensive surveys of genome occupancy, greater resolution, and lower cost for whole genome coverage. Herein, we provide a comparison of high-throughput sequencing platforms and a survey of ChIP-seq analysis tools, discuss experimental design, and describe a detailed ChIP-seq method.Chromatin immunoprecipitation (ChIP) allows enrichment of genomic regions which are associated with specific transcription factors, histone modifications, and indeed any other epitopes which are present on chromatin. The original ChIP methods used site-specific PCR and Southern blotting to confirm which regions of the genome were enriched, on a candidate basis. The combination of ChIP with genomic tiling arrays (ChIP-chip) allowed a more unbiased approach to map ChIP-enriched sites. However, limitations of microarray probe design and probe number have a detrimental impact on the coverage, resolution, sensitivity, and cost of whole-genome tiling microarray sets for higher eukaryotes with large genomes. The combination of ChIP with high-throughput sequencing technology has allowed more comprehensive surveys of genome occupancy, greater resolution, and lower cost for whole genome coverage. Herein, we provide a comparison of high-throughput sequencing platforms and a survey of ChIP-seq analysis tools, discuss experimental design, and describe a detailed ChIP-seq method.

In vivo protein-DNA interactions at human DNA replication origin.

Protein-DNA interactions were studied in vivo at the region containing a human DNA replication origin, located at the 3' end of the lamin B2 gene and partially overlapping the promoter of another gene, located downstream. DNase I treatment of nuclei isolated from both exponentially growing and nonproliferating HL-60 cells showed that this region has an altered, highly accessible, chromatin structure. High-resolution analysis of protein-DNA interactions in a 600-bp area encompassing the origin was carried out by the in vivo footprinting technique based on the ligation-mediated polymerase chain reaction. In growing HL-60 cells, footprints at sequences homologous to binding sites for known transcription factors (members of the basic-helix-loop-helix family, nuclear respiratory factor 1, transcription factor Sp1, and upstream binding factor) were detected in the region corresponding to the promoter of the downstream gene. Upon conversion of cells to a nonproliferative state, a reduction in the intensity of these footprints was observed that paralleled the diminished transcriptional activity of the genomic area. In addition to these protections, in close correspondence to the replication initiation site, a prominent footprint was detected that extended over 70 nucleotides on one strand only. This footprint was absent from nonproliferating HL-60 cells, indicating that this specific protein-DNA interaction might be involved in the process of origin activation.

Fluorescent microplate-based analysis of protein-DNA interactions I:immobilized protein

A simple protein-DNA interaction analysis has been developed using a high-affinity/high-specificity zinc finger protein. In essence, purified protein samples are immobilized directly onto the surface of microplate wells, and fluorescently labeled DNA is added in solution. After incubation and washing, bound DNA is detected in a standard microplate reader. The minimum sensitivity of the assay is approximately 0.2 nM DNA. Since the detection of bound DNA is noninvasive and the protein-DNA interaction is not disrupted during detection, iterative readings may be taken from the same well, after successive alterations in interaction conditions, if required. In this respect, the assay may therefore be considered real time and permits appropriate interaction conditions to be determined quantitatively. The assay format is ideally suited to investigate the interactions of purified unlabeled DNA binding proteins in a high-throughput format.

Fluorescent microplate-based analysis of protein-DNA interactions II:immobilized DNA

A simple protein-DNA interaction analysis has been developed using both a high-affinity/high-specificity zinc finger protein and a low-specificity zinc finger protein with nonspecific DNA binding capability. The latter protein is designed to mimic background binding by proteins generated in randomized or shuffled gene libraries. In essence, DNA is immobilized onto the surface of microplate wells via streptavidin capture, and green fluorescent protein (GFP)-labeled protein is added in solution as part of a crude cell lysate or protein mixture. After incubation and washing, bound protein is detected in a standard microplate reader. The minimum sensitivity of the assay is approximately 0.4 nM protein. The assay format is ideally suited to investigate the interactions of DNA binding proteins from within crude cell extracts and/or mixtures of proteins that may be encountered in protein libraries generated by codon randomization or gene shuffling.

Mapping Protein-DNA Interactions Using ChIP-exo and Illumina-Based Sequencing

Chromatin immunoprecipitation (ChIP) provides a means of enriching DNA associated with transcription factors, histone modifications, and indeed any other proteins for which suitably characterized antibodies are available. Over the years, sequence detection has progressed from quantitative real-time PCR and Southern blotting to microarrays (ChIP-chip) and now high-throughput sequencing (ChIP-seq). This progression has vastly increased the sequence coverage and data volumes generated. This in turn has enabled informaticians to predict the identity of multi-protein complexes on DNA based on the overrepresentation of sequence motifs in DNA enriched by ChIP with a single antibody against a single protein. In the course of the development of high-throughput sequencing, little has changed in the ChIP methodology until recently. In the last three years, a number of modifications have been made to the ChIP protocol with the goal of enhancing the sensitivity of the method and further reducing the levels of nonspecific background sequences in ChIPped samples. In this chapter, we provide a brief commentary on these methodological changes and describe a detailed ChIP-exo method able to generate narrower peaks and greater peak coverage from ChIPped material.

Molecular dynamics studies of the STAT3 homodimer:DNA complex: relationships between STAT3 mutations and protein-DNA recognition.

Signal Transducers and Activators of Transcription (STAT) proteins are a group of latent cytoplasmic transcription factors involved in cytokine signaling. STAT3 is a member of the STAT family and is expressed at elevated levels in a large number of diverse human cancers and is now a validated target for anticancer drug discovery.. Understanding the dynamics of the STAT3 dimer interface, accounting for both protein-DNA and protein-protein interactions, with respect to the dynamics of the latent unphosphorylated STAT3 monomer, is important for designing potential small-molecule inhibitors of the activated dimer. Molecular dynamics (MD) simulations have been used to study the activated STAT3 homodimer:DNA complex and the latent unphosphorylated STAT3 monomer in an explicit water environment. Analysis of the data obtained from MD simulations over a 50 ns time frame has suggested how the transcription factor interacts with DNA, the nature of the conformational changes, and ways in which function may be affected. Examination of the dimer interface, focusing on the protein-DNA interactions, including involvement of water molecules, has revealed the key residues contributing to the recognition events involved in STAT3 protein-DNA interactions. This has shown that the majority of mutations in the DNA-binding domain are found at the protein-DNA interface. These mutations have been mapped in detail and related to specific protein-DNA contacts. Their structural stability is described, together with an analysis of the model as a starting-point for the discovery of novel small-molecule STAT3 inhibitors.

Chemical studies of viral entry mechanisms: I. Hydrophobic protein-lipid interactions during sendai virus membrane fusion. II. Kinetics of bacteriophage λ DNA injection.

Viruses possess very specific methods of targeting and entering cells. These methods would be extremely useful if they could also be applied to drug delivery, but little is known about the molecular mechanisms of the viral entry process. In order to gain further insight into mechanisms of viral entry, chemical and spectroscopic studies in two systems were conducted, examining hydrophobic protein-lipid interactions during Sendai virus membrane fusion, and the kinetics of bacteriophage λ DNA injection.

Sendai virus glycoprotein interactions with target membranes during the early stages of fusion were examined using time-resolved hydrophobic photoaffinity labeling with the lipid-soluble carbene generator3-(trifluoromethyl)-3-(m-^(125 )I] iodophenyl)diazirine (TID). The probe was incorporated in target membranes prior to virus addition and photolysis. During Sendai virus fusion with liposomes composed of cardiolipin (CL) or phosphatidylserine (PS), the viral fusion (F) protein is preferentially labeled at early time points, supporting the hypothesis that hydrophobic interaction of the fusion peptide at the N-terminus of the F_1 subunit with the target membrane is an initiating event in fusion. Correlation of the hydrophobic interactions with independently monitored fusion kinetics further supports this conclusion. Separation of proteins after labeling shows that the F_1 subunit, containing the putative hydrophobic fusion sequence, is exclusively labeled, and that the F_2 subunit does not participate in fusion. Labeling shows temperature and pH dependence consistent with a need for protein conformational mobility and fusion at neutral pH. Higher amounts of labeling during fusion with CL vesicles than during virus-PS vesicle fusion reflects membrane packing regulation of peptide insertion into target membranes. Labeling of the viral hemagglutinin/neuraminidase (HN) at low pH indicates that HN-mediated fusion is triggered by hydrophobic interactions, after titration of acidic amino acids. HN labeling under nonfusogenic conditions reveals that viral binding may involve hydrophobic as well as electrostatic interactions. Controls for diffusional labeling exclude a major contribution from this source. Labeling during reconstituted Sendai virus envelope-liposome fusion shows that functional reconstitution involves protein retention of the ability to undergo hydrophobic interactions.

Examination of Sendai virus fusion with erythrocyte membranes indicates that hydrophobic interactions also trigger fusion between biological membranes, and that HN binding may involve hydrophobic interactions as well. Labeling of the erythrocyte membranes revealed close membrane association of spectrin, which may play a role in regulating membrane fusion. The data show that hydrophobic fusion protein interaction with both artificial and biological membranes is a triggering event in fusion. Correlation of these results with earlier studies of membrane hydration and fusion kinetics provides a more detailed view of the mechanism of fusion.

The kinetics of DNA injection by bacteriophage λ. into liposomes bearing reconstituted receptors were measured using fluorescence spectroscopy. LamB, the bacteriophage receptor, was extracted from bacteria and reconstituted into liposomes by detergent removal dialysis. The DNA binding fluorophore ethidium bromide was encapsulated in the liposomes during dialysis. Enhanced fluorescence of ethidium bromide upon binding to injected DNA was monitored, and showed that injection is a rapid, one-step process. The bimolecular rate law, determined by the method of initial rates, revealed that injection occurs several times faster than indicated by earlier studies employing indirect assays.

It is hoped that these studies will increase the understanding of the mechanisms of virus entry into cells, and to facilitate the development of virus-mimetic drug delivery strategies.

Quantitative analysis of DNA-protein interactions using double-labeled native gel electrophoresis and fluorescence-based imaging

We have developed a sensitive, non-radioactive method to assess the interaction of transcription factors/DNA-binding proteins with DNA. We have modified the traditional radiolabeled DNA gel mobility shift assay to incorporate a DNA probe end-labeled with a Texas-red fluorophore and a DNA-binding protein tagged with the green fluorescent protein to monitor precisely DNA-protein complexation by native gel electrophoresis. We have applied this method to the DNA-binding proteins telomere release factor-1 and the sex-determining region-Y, demonstrating that the method is sensitive (able to detect 100 fmol of fluorescently labeled DNA), permits direct visualization of both the DNA probe and the DNA-binding protein, and enables quantitative analysis of DNA and protein complexation, and thereby an estimation of the stoichiometry of protein-DNA binding.

Quantum clustering and network analysis of MD simulation trajectories to probe the conformational ensembles of protein-ligand interactions

In this article, we present a novel application of a quantum clustering (QC) technique to objectively cluster the conformations, sampled by molecular dynamics simulations performed on different ligand bound structures of the protein. We further portray each conformational population in terms of dynamically stable network parameters which beautifully capture the ligand induced variations in the ensemble in atomistic detail. The conformational populations thus identified by the QC method and verified by network parameters are evaluated for different ligand bound states of the protein pyrrolysyl-tRNA synthetase (DhPylRS) from D. hafniense. The ligand/environment induced re-distribution of protein conformational ensembles forms the basis for understanding several important biological phenomena such as allostery and enzyme catalysis. The atomistic level characterization of each population in the conformational ensemble in terms of the re-orchestrated networks of amino acids is a challenging problem, especially when the changes are minimal at the backbone level. Here we demonstrate that the QC method is sensitive to such subtle changes and is able to cluster MD snapshots which are similar at the side-chain interaction level. Although we have applied these methods on simulation trajectories of a modest time scale (20 ns each), we emphasize that our methodology provides a general approach towards an objective clustering of large-scale MD simulation data and may be applied to probe multistate equilibria at higher time scales, and to problems related to protein folding for any protein or protein-protein/RNA/DNA complex of interest with a known structure.

Computational design of self-assembling proteins and protein-DNA nanowires

Computational protein design (CPD) is a burgeoning field that uses a physical-chemical or knowledge-based scoring function to create protein variants with new or improved properties. This exciting approach has recently been used to generate proteins with entirely new functions, ones that are not observed in naturally occurring proteins. For example, several enzymes were designed to catalyze reactions that are not in the repertoire of any known natural enzyme. In these designs, novel catalytic activity was built de novo (from scratch) into a previously inert protein scaffold. In addition to de novo enzyme design, the computational design of protein-protein interactions can also be used to create novel functionality, such as neutralization of influenza. Our goal here was to design a protein that can self-assemble with DNA into nanowires. We used computational tools to homodimerize a transcription factor that binds a specific sequence of double-stranded DNA. We arranged the protein-protein and protein-DNA binding sites so that the self-assembly could occur in a linear fashion to generate nanowires. Upon mixing our designed protein homodimer with the double-stranded DNA, the molecules immediately self-assembled into nanowires. This nanowire topology was confirmed using atomic force microscopy. Co-crystal structure showed that the nanowire is assembled via the desired interactions. To the best of our knowledge, this is the first example of a protein-DNA self-assembly that does not rely on covalent interactions. We anticipate that this new material will stimulate further interest in the development of advanced biomaterials.

The Xanthomonas citri effector protein PthA interacts with citrus proteins involved in nuclear transport, protein folding and ubiquitination associated with DNA repair

P>Xanthomonas axonopodis pv. citri utilizes the type III effector protein PthA to modulate host transcription to promote citrus canker. PthA proteins belong to the AvrBs3/PthA family and carry a domain comprising tandem repeats of 34 amino acids that mediates protein-protein and protein-DNA interactions. We show here that variants of PthAs from a single bacterial strain localize to the nucleus of plant cells and form homo- and heterodimers through the association of their repeat regions. We hypothesize that the PthA variants might also interact with distinct host targets. Here, in addition to the interaction with alpha-importin, known to mediate the nuclear import of AvrBs3, we describe new interactions of PthAs with citrus proteins involved in protein folding and K63-linked ubiquitination. PthAs 2 and 3 preferentially interact with a citrus cyclophilin (Cyp) and with TDX, a tetratricopeptide domain-containing thioredoxin. In addition, PthAs 2 and 3, but not 1 and 4, interact with the ubiquitin-conjugating enzyme complex formed by Ubc13 and ubiquitin-conjugating enzyme variant (Uev), required for K63-linked ubiquitination and DNA repair. We show that Cyp, TDX and Uev interact with each other, and that Cyp and Uev localize to the nucleus of plant cells. Furthermore, the citrus Ubc13 and Uev proteins complement the DNA repair phenotype of the yeast Delta ubc13 and Delta mms2/uev1a mutants, strongly indicating that they are also involved in K63-linked ubiquitination and DNA repair. Notably, PthA 2 affects the growth of yeast cells in the presence of a DNA damage agent, suggesting that it inhibits K63-linked ubiquitination required for DNA repair.

AN ANALYSIS OF THE DNA DAMAGE INDUCED BY NICKEL COMPOUNDS IN CHINESE HAMSTER OVARY CELLS (CARCINOGENESIS, HETEROCHROMATIN, CYTOTOXICITY, REPLICATION, PROTEIN CROSSLINKING)

The carcinogenic activity of water-insoluble crystalline nickel sulfide requires phagocytosis and lysosome-mediated intracellular dissolution of the particles to yield Ni('2+). This study investigated the extent and nature of the DNA damage in Chinese hamster ovary cells treated with various nickel compounds using the technique of alkaline elution. Crystalline NiS and water-soluble NiCl(,2) induced single strand breaks that were repaired quickly and DNA-protein crosslinks that persisted up to 24 hr after exposure to nickel. The induction of single strand breaks was concentration dependent at both noncytotoxic and lethal amounts of nickel. The induction of DNA-protein crosslinks was concentration dependent but was absent at lethal amounts of nickel. The cytoplasmic and nuclear uptake of nickel was concentration dependent even at the toxic level of nickel. However, the induction of DNA-protein crosslinks by nickel required active cell cycling and occurred predominantly in mid-late S phase of the cell cycle, suggesting that the lethal amounts of nickel inhibited DNA-protein crosslinking by inhibiting active cell cycling. Since the DNA-protein crosslinking induced by nickel was resistant to DNA repair, the nature of this lesion was investigated using various methods of DNA isolation and chromatin fractionation in combination with SDS-polyacrylamide gel electrophoresis. High molecular weight, non-histone chromosomal proteins and possibly histone 1 were preferentially crosslinked to DNA by nickel. The crosslinked proteins were concentrated in a magnesium-insoluble fraction of sonicated chromatin (5% of the total) that was similar to heterochromatin in solubility and protein composition. Alterations in DNA structure and function, brought about by the effect of nickel on protein-DNA interactions, may be related to the carcinogenicity of nickel compounds. ^

High-resolution mapping of nucleoprotein complexes by site-specific protein-DNA photocrosslinking: organization of the human TBP-TFIIA-TFIIB-DNA quaternary complex.

We have used a novel site-specific protein-DNA photocrosslinking procedure to define the positions of polypeptide chains relative to promoter DNA in binary, ternary, and quaternary complexes containing human TATA-binding protein, human or yeast transcription factor IIA (TFIIA), human transcription factor IIB (TFIIB), and promoter DNA. The results indicate that TFIIA and TFIIB make more extensive interactions with promoter DNA than previously anticipated. TATA-binding protein, TFIIA, and TFIIB surround promoter DNA for two turns of DNA helix and thus may form a "cylindrical clamp" effectively topologically linked to promoter DNA. Our results have implications for the energetics, DNA-sequence-specificity, and pathway of assembly of eukaryotic transcription complexes.