UNSUPERVISED INDEXING OF MEDLINE ARTICLES THROUGH GRAPH-BASED RANKING

The biomedical literature is extensively catalogued and indexed in MEDLINE. MEDLINE indexing is done by trained human indexers, who identify the most important concepts in each article, and is expensive and inconsistent. Automating the indexing task is difficult: the National Library of Medicine produces the Medical Text Indexer (MTI), which suggests potential indexing terms to the indexers. MTI’s output is not good enough to work unattended. In my thesis, I propose a different way to approach the indexing task called MEDRank. MEDRank creates graphs representing the concepts in biomedical articles and their relationships within the text, and applies graph-based ranking algorithms to identify the most important concepts in each article. I evaluate the performance of several automated indexing solutions, including my own, by comparing their output to the indexing terms selected by the human indexers. MEDRank outperformed all other evaluated indexing solutions, including MTI, in general indexing performance and precision. MEDRank can be used to cluster documents, index any kind of biomedical text with standard vocabularies, or could become part of MTI itself.

Loss of DNA polymerase zeta enhances spontaneous tumorigenesis.

Mammalian genomes encode at least 15 distinct DNA polymerases, functioning as specialists in DNA replication, DNA repair, recombination, or bypass of DNA damage. Although the DNA polymerase zeta (polzeta) catalytic subunit REV3L is important in defense against genotoxins, little is known of its biological function. This is because REV3L is essential during embryogenesis, unlike other translesion DNA polymerases. Outstanding questions include whether any adult cells are viable in the absence of polzeta and whether polzeta status influences tumorigenesis. REV3L-deficient cells have properties that could influence the development of neoplasia in opposing ways: markedly reduced damage-induced point mutagenesis and extensive chromosome instability. To answer these questions, Rev3L was conditionally deleted from tissues of adult mice using MMTV-Cre. Loss of REV3L was tolerated in epithelial tissues but not in the hematopoietic lineage. Thymic lymphomas in Tp53(-/-) Rev3L conditional mice occurred with decreased latency and higher incidence. The lymphomas were populated predominantly by Rev3L-null T cells, showing that loss of Rev3L can promote tumorigenesis. Remarkably, the tumors were frequently oligoclonal, consistent with accelerated genetic changes in the absence of Rev3L. Mammary tumors could also arise from Rev3L-deleted cells in both Tp53(+/+) and Tp53(+/-) backgrounds. Mammary tumors in Tp53(+/-) mice deleting Rev3L formed months earlier than mammary tumors in Tp53(+/-) control mice. Prominent preneoplastic changes in glandular tissue adjacent to these tumors occurred only in mice deleting Rev3L and were associated with increased tumor multiplicity. Polzeta is the only specialized DNA polymerase yet identified that inhibits spontaneous tumor development.

A molecular analysis of the conditional defect in MuSVts 110 viral RNA splicing

Cells infected with MuSVts110 express a viral RNA which contains an inherent conditional defect in RNA splicing. It has been shown previously that splicing of the MuSVts110 primary transcript is essential to morphological transformation of 6m2 cells in vitro. A growth temperature of 33$\sp\circ$C is permissive for viral RNA splicing,and, consequently, 6m2 cells appear morphologically transformed at this temperature. However, 6m2 cells appear phenotypically normal when incubated at 39$\sp\circ$C, the non-permissive temperature for viral RNA splicing.^ After a shift from 39$\sp\circ$C to 33$\sp\circ$C, the coordinate splicing of previously synthesized and newly transcribed MuSVts110 RNA was achieved. By S1 nuclease analysis of total RNA isolated at various times, 5$\sp\prime$ splice site cleavage of the MuSVts110 transcript appeared to occur 60 minutes after the shift to 33$\sp\circ$C, and 30 minutes prior to detectable exon ligation. In addition, consistent with the permissive temperatures and the kinetic timeframe of viral RNA splicing after a shift to 33$\sp\circ$C, four temperature sensitive blockades to primer extension were identified 26-75 bases upstream of the 3$\sp\prime$ splice site. These blockades likely reflect four branchpoint sequences utilized in the formation of MuSVts110 lariat splicing-intermediates.^ The 54-5A4 cell line is a spontaneous revertant of 6m2 cells and appears transformed at all growth temperatures. Primer extension sequence analysis has shown that a five base deletion occurred at the 3$\sp\prime$ splice site in MuSVts110 RNA allowing the expression of a viral transforming protein in 54-5A4 in the absence of RNA splicing, whereas in the parental 6m2 cell line, a splicing event is necessary to generate a similar transforming protein. As a consequence of this deletion, splicing cannot occur and the formation of the four MuSVts110 branched-intermediates were not observed at any temperature in 54-5A4 cells. However, 5$\sp\prime$ splice site cleavage was still detected at 33$\sp\circ$C.^ Finally, we have investigated the role of the 1488 bp deletion which occurred in the generation of MuSVts110 in the activation of temperature sensitive viral RNA splicing. This deletion appears solely responsible for splice site activation. Whether intron size is the crucial factor in MuSVts110 RNA splicing or whether inhibitory sequences were removed by the deletion is currently unknown. (Abstract shortened with permission of author.) ^

Immunochemical interactions of peptides with site-directed polyclonal immunoglobulin G

The objective of this study was to investigate the immunochemical nature of the polyclonal immune response to the 14mer peptide TINKEDDESPGLYG and to identify interactions among antibodies to more than one epitope. Two groups of rabbits were immunized with the 14mer peptide and a Keyhole Limpet hemocyanin (KLH) carrier, but with KLH attached either to the 14mer's N- or C-terminus. Two approximate epitopes were mapped by an antibody-capture enzyme-linked immunosorbent assay method using antiserum obtained when KLH was oriented on the C-terminus of the 14mer. A precise mapping of the epitopes performed with inhibition enzyme immunoassays (iEIAs) resulted in an N-terminal 6mer epitope TINKED and a C-terminal 10mer epitope EDDESPGLYG. The epitopes overlapped by two amino acids. IEIAs and iEIAs incorporating antibody-blocking peptides indicated that the two anti-epitope antibody fractions did not interfere with one anothers' epitope binding. It was postulated that the anti-TINKED and anti-EDDESPGLYG antibody fractions individually bind their respective hydrophobic epitope "core" region at the N- or C-terminal of peptide TINKEDDESPGLYG, while sharing the two hydrophilic overlap amino acids. This antibody "lap joint" binding interaction can be accomplished by each of the anti-epitope antibodies binding an opposite side of the epitope overlap region in the shallow periphery of its binding site. ^

Analysis of BMP signalling through the receptor type IA in embryogenesis using conditional gene inactivation in mice

Although bone morphogenetic proteins (BMPs) were initially identified for their potent bone-inducing activity, their precise roles in processes of endochondral and intramembranous bone formation are far from being clear. Tissue-specific loss-of-function experiments using the BMP receptor type IA (BMPR-IA) are particularly attractive since this receptor is thought to be essential for signaling by the closely related BMPs -2, 4, and 7. To ablate signaling through this receptor during chondrogenesis, we have generated transgenic mice expressing Cre recombinase under the control of the collagen type II (Col2a1) gene regulatory sequences. Mice lacking BMPR-IA function in chondrocytes display a number of skeletal abnormalities, including defects in bones of the chondrocranium, abnormal dorsal vertebral processes, scapulae with severe hypoplasia of dorsal elements, and shortening of the long bones. Alterations in the growth plate of long bones in mutants suggest that BMPR-IA is not required for early steps of the chondrocyte specification, but is rather important in regulation of terminal differentiation. Molecular analysis revealed noticeable downregulation of the Ihh/Ptch signalling pathway, decreased chondrocyte proliferation rate and deregulation of hypertrophy. ^ In order to elucidate the role of BMP signalling in development of the limb and intramembranous ossification, we have used mice expressing Cre recombinase under control of the Prx1 (MHox) regulatory elements (M. Logan, pers comm.). Cre activity was found in those mice in the developing limb bud mesenchyme, as well as in a subset of cranial neural crest cells. Prx1-Cre-induced conditional mutants display prominent defects in distal limb outgrowth, as well as ossification defects in a number of neural crest-derived calvarial bones. Intriguingly, mutant limbs displayed alterations in patterning along all three axes. Molecular analysis revealed ectopic anterior Shh/Ptch signalling pathway activation and expression of some Hox genes. Observed loss of Msx1 and Msx2 expression in the progress zone correlates with downregulation of Cyclin D1 and decreased distal outgrowth. Abnormal ventral localization of Lmx1b-expressing cells along with observed later morphological abnormalities suggest a novel role for BMP signalling in establishment or maintaining of the dorso-ventral polarity in the limb mesoderm. ^

Conditional expression of Sry, and Wnt4 by gene-targeting: Studies in sexual and skeletal development

Sry and Wnt4 cDNAs were individually introduced into the ubiquitously-expressed Rosa26 ( R26) locus by gene targeting in embryonic stem (ES) cells to create a conditional gene expression system in mice. In the targeted alleles, expression of these cDNAs should be blocked by a neomycin resistance selection cassette that is flanked by loxP sites. Transgene expression should be activated after the blocking cassette is deleted by Cre recombinase. ^ To test this conditional expression system, I have bred R26-stop- Sry and R26-stop-Wnt4 heterozygotes with a MisRII-Cre mouse line that expresses Cre in the gonads of both sexes. Analysis of these two types of bigenic heterozygotes indicated that their gonads developed normally like those of wild types. However, one XX R26-Sry/R26-Sry; MisR2-Cre/+ showed epididymis-like structures resembling those of males. In contrast, only normal phenotypes were observed in XY R26-Wnt4/R26-Wnt4; MisR2-Cre /+ mice. To interpret these results, I have tested for Cre recombinase activity by Southern blot and transcription of the Sry and Wnt4 transgenes by RT-PCR. Results showed that bigenic mutants had insufficient activation of the transgenes in their gonads at E12.5 and E13.5. Therefore, the failure to observe mutant phenotypes may have resulted from low activity of MisR2-Cre recombination at the appropriate time. ^ Col2a1-Cre transgenic mice express Cre in differentiating chondrocytes. R26-Wnt4; Col2a1-Cre bigenic heterozygous mice were found to exhibit a dramatic alteration in growth presumably caused by Wnt4 overexpression during chondrogenesis. R26-Wnt4; Col2a1-Cre mice exhibited dwarfism beginning approximately 10 days after birth. In addition, they also had craniofacial abnormalities, and had delayed ossification of the lumbar vertebrate and pelvic bones. Histological analysis of the growth plates of R26-Wnt4; Col2a1-Cre mice revealed less structural organization and a delay in onset of the primary and secondary ossification centers. Molecular studies confirmed that overexpression of Wnt4 causes decreased proliferation and early maturation of chondrocytes. In addition, R26-Wnt4; Col2a1-Cre mice had decreased expression of vascular endothelial growth factor (VEGF), suggesting that defects in vascularization may contribute to the dwarf phenotype. Finally, 9-month-old R26-Wnt4; Col2a1-Cre mice had significantly more fat cells in the marrow cavities of their metaphysis long bones, implying that long-term overexpression of Wnt4may cause bone marrow pathologies. In conclusion, Wnt4 was activated by Col2a1-Cre recombinase and was overexpressed in the growth plate, resulting in aberrant proliferation and differentiation of chondrocytes, and ultimately leads to dwarfism in mice. ^

Analyzing the function of myogenin in adult satellite cells through the construction of a myogenin conditional allele

Although mechanisms regulating the formation of embryonic skeletal muscle are well characterized, less is known about muscle formation in postnatal life. This disparity is unfortunate because the largest increases in skeletal muscle mass occur after birth. Adult muscle stem cells (satellite cells) appear to recapitulate the events that occur in embryonic myoblasts. In particular, the myogenic basic helix-loop-helix factors, which have crucial functions in embryonic muscle development, are assumed to have similar roles in postnatal muscle formation. Here, I test this assumption by determining the role of the myogenic regulator myogenin in postnatal life. Myogenin-null mice die at birth, necessitating the generation of floxed alleles of myogenin and the use of cre-recombinase lines to delete myogenin. Removing myogenin before embryonic muscle development resulted in myofiber deficiencies identical to those observed in myogenin-null mice. However, mice in which myogenin was deleted following embryonic muscle development had normal skeletal muscle, except for modest alterations in MRF4 and MyoD expression. Notably, myogenin-deleted mice were 30% smaller than controls, suggesting that myogenin's absence disrupted general body growth. These results suggest that skeletal muscle growth in postnatal life is controlled by mechanisms distinct from those occurring in embryonic muscle development. ^

A conditional Mdm2 allele shows the importance of Mdm2 in heart development

Over 50% of sporadic tumors in humans have a p53 mutation highlighting its importance as a tumor suppressor. Considering additional mutations in other genes involved in p53 pathways, every tumor probably has mutant p53 or impaired p53-mediated functions. In response to a variety of cellular and genotoxic stresses, p53, mainly through its transcriptional activity, induces pathways involved in apoptosis and growth arrest. In these circumstances and under normal situations, p53 must be tightly regulated. Mdm2 is an important regulator of p53. Mdm2 inhibits p53 function by binding and blocking its transactivation domain. In addition, Mdm2 helps target p53 for degradation through its E3 ligase activity. Mdm2 null mice are embryonic lethal due to apoptosis in the blastocysts. However, a p53 null background rescues this lethality demonstrating the importance of the p53-Mdm2 interaction, particularly during development. The lethality of the Mdm2 null mouse prior to implantation limits the ability to investigate the role of Mdm2 in regulating p53 in a temporal and tissue specific manner. Does p53 need to be regulated in all tissues throughout the life of a mouse? Does Mdm2 always have to regulate it? To address these questions, we created a conditional Mdm2 allele. The conditional allele, Mdm2FM, in the presence of Cre recombinase results in the deletion of exons 5 and 6 of Mdm2 (most of the p53 binding domain) and represents a null allele. ^ The Mdm2FM allele was crossed with a heart muscle specific Cre expressing mouse (α-myosin heavy chain promoter driven Cre) to ask whether Mdm2 acts as a negative regulator of p53 in the heart. The heart is the most prominent organ early in embryogenesis and is shaped by cell death and proliferation. p53 does not appear to be active in the heart in response to some types of stress, so it remained to be determined if it has to be regulated in normal heart development. Loss of Mdm2 in the heart results in heart defects as early as E9.5. Loss of Mdm2 results in stabilized p53 and apoptosis. This apoptosis leads to a thinning of the myocardial wall particularly in the ventricles and abnormal ventricular structure. Eventually the abnormal heart fails resulting in lethality by E13.5. The embryonic lethality is rescued in a p53 null background. Thus, Mdm2 is important in regulating p53 in the development of the heart. ^

Ligand-directed targeting of malignant gliomas

The dramatic poor survival of patients diagnosed with glioblastoma multiforme (GBM) is a reflection of the struggles that accompany traditional treatments. Thus, the development of molecular-based targeted therapies represents new windows for intervention. In this study, we hypothesized that we could select peptide-ligands that selectively target GBM based on the idea that the glioma microenvironment may induce or modify the expression of cell surface receptors that could be accessed by circulating peptides. To select the peptides we employed two distinct in vivo screenings. First, a random phage-displayed peptide library was injected into mice bearing intracranial tumors. Phage that bound to tumor were recovered and sequenced. We found that the tumor-derived phage CLSYKGRC, CNKVSTKC and CQSSREKC were recovered with the highest frequencies and used for subsequent targeting experiments. Second, the phage peptide library was injected into mice without tumors and phage were recovered from brain and sequenced. A phage-displayed peptide (CRTIGPSVC) with homology to transferrin (Tf) was selected and injected into brain tumor-bearing mice. Results showed that after 6 hours of circulation, the CLSYKGRC, CNKVSTKC and CQSSREKC-phage selectively targeted GBM vasculature. In contrast, Tf-like phage accumulated outside the tumor blood vessels in the cytoplasm of cells located within GBM, suggesting it was internalized in vivo. However, after short periods of circulation this phage was restricted to the tumor vasculature. Importantly, none of the selected phage targeted normal brain cells in animals bearing intracranial tumors. An affinity column coupled to the CNKVSTKC zpeptide was used to identify receptors from GBM. Using mass-spectrometry Vimentin, a marker of glial malignancy, was identified as a potential receptor. Other studies showed that the Tf-like phage bound selectively to Apo-Tf (iron free), with no binding to Holo-Tf (iron loaded) or to Tf receptor (TfR). However, the binding of Tf-like phage to glioma cells that express TfR increased in the presence of Apo-Tf. Thus, the Tf-like phage could indirectly target TfR using the endogenous Tf pathway. We propose that the novel peptides identified in this study could be conjugated to therapeutic or imaging agents for use GBM. ^

Bcr: A conditional inhibitor of Bcr -Abl oncoprotein

Chronic myelogenous leukemia (CML) is characterized cytogenetically by the presence of the Philadelphia chromosome and clinically by the clonal expansion of the hematopoietic stem cells and the accumulation of large numbers of myeloid cells. Philadelphia chromosome results from the reciprocal translocation between chromosomes 9 and 22 [t(9;22)(324;q11)], which fuses parts of the ABL proto-oncogene to 5′ portions of the BCR gene. The product of the fused gene is Bcr-Abl oncoprotein. Bcr-Abl oncoprotein has elevated protein tyrosine kinase activity, and is the cause of Philadelphia chromosome associated leukemias. The Bcr sequence in the fusion protein is crucial for the activation of Abl kinase activity and transforming phenotype of Bcr-Abl oncoprotein. Although the Bcr-Abl oncoprotein has been studied extensively, its normal counterpart, the Bcr protein, has been less studied and its function is not well understood. At this point, Bcr is known to encode a novel serine/threonine protein kinase. In Bcr-Abl positive leukemia cells, we found that the serine kinase activity of Bcr is impaired by tyrosine phosphorylation. Both the Bcr protein sequences within Bcr-Abl and the normal cellular Bcr protein lack serine/threonine kinase activity when they become phosphorylated on tyrosine residues by Bcr-Abl. Therefore, the goal of this study was to investigate the role of Bcr in Bcr-Abl positive leukemia cells. We found that overexpression of Bcr can inhibit Bcr-Abl tyrosine kinase activity, and the inhibition is dependent on its intact serine/threonine kinase function. Using the tet repressible promoter system, we demonstrated that Bcr when induced in Bcr-Abl positive leukemia cells inhibited the Bcr-Abl oncoprotein tyrosine kinase. Furthermore, induction of Bcr also increased the number of cells undergoing apoptosis and inhibited the transforming ability of Bcr-Abl. In contrast to the wild-type Bcr, the kinase-inactive mutant of Bcr (Y328F/Y360F) had no effects on Bcr-Abl tyrosine kinase in cells. Results from other experiments indicated that phosphoserine-containing Bcr sequences within the first exon, which are known to bind to the Abl SH2 domain, are responsible for observed inhibition of the Bcr-Abl tyrosine kinase. Several lines of evidence suggest that the phosphoserine form of Bcr, which binds to the Abl SH2 domain, strongly inhibits the Abl tyrosine kinase domain of Bcr-Abl Previously published findings from our laboratory have also shown that Bcr is phosphorylated on tyrosine residue 177 in Bcr-Abl positive cells and that this form of Bcr recruits the Grb2 adaptor protein, which is known to activate the Ras pathway. These findings implicate Bcr as an effector of Bcr-Abl's oncogenic activity. Therefore based on the findings presented above, we propose a model for dual Function of Bcr in Bcr-Abl positive leukemia cells. Bcr, when active as a serine/threonine kinase and thus autophosphorylating its own serine residues, inhibits Bcr-Abl's oncogenic functions. However, when Ber is tyrosine phosphorylated, its Bcr-Abl inhibitory function is neutralized thus allowing Bcr-Abl to exert its full oncogenic potential. Moreover, tyrosine phosphorylated Bcr would compliment Bcr-Abl's neoplastic effects by the activation of the Ras signaling pathway. ^

Circumvention of cellular defense responses to DNA -directed nucleoside analogues by the protein kinase inhibitor, UCN-01

DNA-directed nucleoside analogues, such as ara-C, fludarabine, and gemcitabine, are antimetabolites effective in the treatment of a variety of cancers. However, resistance to nucleoside analogue-based chemotherapy in treatments is still a major problem in therapy. Therefore, it is essential to develop rationales for optimizing the use of nucleoside analogues in combination with other anticancer drugs or modalities such as radiation. The present study focuses on establishing mechanism-based combination strategy to overcome resistance to nucleoside analogues. ^ I hypothesized that the cytostatic concentrations of nucleoside analogues may cause S-phase arrest by activating an S-phase checkpoint that consists of a series of kinases. This may allow cells to repair damaged DNA over time and spare cytotoxicity. Thus, the ability of cells to enact an S-phase arrest in response to incorporation of potentially lethal amounts of nucleoside analogue may serve as a mechanism of resistance to S-phase-specific agents. As a corollary, the addition of a kinase inhibitor, such as UCN-01, may dysregulate the checkpoint response and abrogate the survival of S-phase-arrested cells by suppression of the survival signaling pathways. Using gemcitabine as a model of S-phase-specific nucleoside analogues in human acute myelogenous leukemia ML-1 cells, I demonstrated that cells arrested in S-phase in response to cytostatic conditions. Proliferation continued after washing the cells into drug-free medium, suggesting S-phase arrest served as a resistance mechanism of cancer cells to spare cytotoxicity of nucleoside analogues. However, nontoxic concentrations of UCN-01 rapidly killed S-phase-arrested cells by apoptosis. Furthermore, the molecular mechanism for UCN-01-induced apoptosis in S-phase-arrested cells was through inhibition of survival pathways associated with these cells. In this regard, suppression of the PI 3-kinase-Akt-Bad survival pathway as well as the NF-κB signaling pathway were associated with induction of apoptosis in S-phase-arrested cells by UCN-01, whereas the Ras-Raf-MEK-ERK pathway appeared not involved. This study has provided the rationales and strategies for optimizing the design of effective combination therapies to overcome resistance to nucleoside analogues. In fact, a clinical trial of the combination of ara-C with UCN-01 to treat relapsed or refractory AML patients has been initiated at U.T.M.D. Anderson Cancer Center. ^