Squared Extrapolation Methods (SQUAREM): A New Class of Simple and Efficient Numerical Schemes for Accelerating the Convergence of the EM Algorithm

We derive a new class of iterative schemes for accelerating the convergence of the EM algorithm, by exploiting the connection between fixed point iterations and extrapolation methods. First, we present a general formulation of one-step iterative schemes, which are obtained by cycling with the extrapolation methods. We, then square the one-step schemes to obtain the new class of methods, which we call SQUAREM. Squaring a one-step iterative scheme is simply applying it twice within each cycle of the extrapolation method. Here we focus on the first order or rank-one extrapolation methods for two reasons, (1) simplicity, and (2) computational efficiency. In particular, we study two first order extrapolation methods, the reduced rank extrapolation (RRE1) and minimal polynomial extrapolation (MPE1). The convergence of the new schemes, both one-step and squared, is non-monotonic with respect to the residual norm. The first order one-step and SQUAREM schemes are linearly convergent, like the EM algorithm but they have a faster rate of convergence. We demonstrate, through five different examples, the effectiveness of the first order SQUAREM schemes, SqRRE1 and SqMPE1, in accelerating the EM algorithm. The SQUAREM schemes are also shown to be vastly superior to their one-step counterparts, RRE1 and MPE1, in terms of computational efficiency. The proposed extrapolation schemes can fail due to the numerical problems of stagnation and near breakdown. We have developed a new hybrid iterative scheme that combines the RRE1 and MPE1 schemes in such a manner that it overcomes both stagnation and near breakdown. The squared first order hybrid scheme, SqHyb1, emerges as the iterative scheme of choice based on our numerical experiments. It combines the fast convergence of the SqMPE1, while avoiding near breakdowns, with the stability of SqRRE1, while avoiding stagnations. The SQUAREM methods can be incorporated very easily into an existing EM algorithm. They only require the basic EM step for their implementation and do not require any other auxiliary quantities such as the complete data log likelihood, and its gradient or hessian. They are an attractive option in problems with a very large number of parameters, and in problems where the statistical model is complex, the EM algorithm is slow and each EM step is computationally demanding.

Neuropathological and molecular comparison between clinical and asymptomatic bovine spongiform encephalopathy cases

Interest in the proper neuropathological and molecular characterization of bovine spongiform encephalopathy (BSE) has increased since asymptomatic and atypical cases were detected in the cattle population by active disease surveillance. In this respect we investigated a total of 95 confirmed BSE cases originating from different active and passive surveillance categories (clinical suspects, emergency-slaughter, fallen stock and routinely slaughter) in Switzerland for their neuropathological and molecular phenotype. We looked for measurable differences between these categories in lesion profile, severity of spongiform change, degree of astrocytosis as well as immunohistochemical and molecular patterns of the disease-associated isoform of the prion protein (PrPd) in the caudal brainstem. Our results indicate significantly higher intensities of spongiform change in clinically affected compared to asymptomatic BSE cases. Similar effects were in trend observed for the intensities of PrPd deposition and astrocytosis, whereas the frequencies of morphological PrPd types and the molecular patterns in Western immunoblot were not different. Importantly, none of the animals included in this study revealed features of atypical BSE. Taken together, this study suggests that both clinically affected as well as asymptomatic Swiss BSE cases in cattle share the neuropathological and molecular phenotype of classical BSE and that asymptomatic classical BSE cases are at a pre-clinical stage of the disease rather than representing a true sub-clinical form of BSE.

Distinct molecular composition of blood and lymphatic vascular endothelial cell junctions establishes specific functional barriers within the peripheral lymph node

Lymph nodes are strategically localized at the interfaces between the blood and lymphatic vascular system, delivering immune cells and antigens to the lymph node. As cellular junctions of endothelial cells actively regulate vascular permeability and cell traffic, we have investigated their molecular composition by performing an extensive immunofluorescence study for adherens and tight junction molecules, including vascular endothelium (VE)-cadherin, the vascular claudins 1, 3, 5 and 12, occludin, members of the junctional adhesion molecule family plus endothelial cell-selective adhesion molecule (ESAM)-1, platelet endothelial cell adhesion molecule-1, ZO-1 and ZO-2. We found that junctions of high endothelial venules (HEV), which serve as entry site for naive lymphocytes, are unique due to their lack of the endothelial cell-specific claudin-5. LYVE-1(+) sinus-lining endothelial cells form a diffusion barrier for soluble molecules that arrive at the afferent lymph and use claudin-5 and ESAM-1 to establish characteristic tight junctions. Analysis of the spatial relationship between the different vascular compartments revealed that HEV extend beyond the paracortex into the medullary sinuses, where they are protected from direct contact with the lymph by sinus-lining endothelial cells. The specific molecular architecture of cellular junctions present in blood and lymphatic vessel endothelium in peripheral lymph nodes establishes distinct barriers controlling the distribution of antigens and immune cells within this tissue.

Simple and efficient synthesis of substituted 2-pyrrolidinones, 2-pyrrolones and pyrrolidines from enaminones of Baylis-Hillman derivatives of 3-isoxazolecarbaldehydes

The enaminones, generated from derivatives of appropriately substituted Baylis-Hillman adducts of 3-isoxazolecarbaldehydes undergo intramolecular ring-closure reactions to afford substituted 2-pyrrolidinones, 1,5-dihydro-2-pyrrolones and N-substituted pyrrolidines in good yields.

Aminoacyl-tRNA synthetases: Probing the molecular basis of catalysis and discrimination

Aminoacyl-tRNA synthetases (RSs) are responsible for the essential connection of amino acids with trinucleotide sequences of tRNA's. The RS family constitutes two structurally dissimilar groups of proteins, class I and class II. Methionyl-tRNA synthetase (MetRS) and isoleucyl-tRNA synthetase (IleRS), both members of class I, were the focus of this work. Both enzymes are zinc-containing proteins; show a high degree of amino acid specificity; and edit activated noncognate amino acids, thereby ensuring the fidelity of the genetic code. The goals of this work were to further delineate the molecular basis of catalysis and discrimination in these enzymes by mapping active site geometries using high-resolution nuclear magnetic resonance spectroscopy (NMR).^ Internuclear distances obtained from transferred nuclear Overhauser effects were used to define the conformations of Mg($\alpha$,$\beta$-methylene)ATP bound to E. coli MetRS and E. coli IleRS in multiple complexes. Identical conformations were found for the bound ATP. Thus, the predicted structural homology between IleRS and MetRS is supported by consensus enzyme-bound nucleotide conformations. The conformation of the bound nucleotide is not sensitive to occupation of the amino acid site of MetRS or IleRS. Therefore, conformational changes known to occur in the synthetases upon ligand binding appear not to alter the bound conformation of the adenosine portion of the nucleotide. Nuclear Overhauser effects on the substrate amino acid L-selenomethionine were also used to evaluate the enzyme-bound conformation of the cognate amino acid. The amino acid assumes a conformation which is consistent with a proposed editing mechanism.^ The E. coli MetRS was shown to catalyze amino acid $\alpha$-proton exchange in the presence of deuterium oxide of all cognate amino acids. It is proposed that the enzyme-bound zinc coordinates the $\alpha$-carboxylate of the amino acid, rendering the $\alpha$-proton more acidic. An enzymic base is responsible for exchange of the $\alpha$-proton. This proposal suggests that the enzyme-bound zinc may have a role in amino acid discrimination in MetRS. However, the role of this exchange reaction in catalysis remains unknown. ^

Mapping and cloning of genes from portions of the human genome using somatic cell hybrids

Human x rodent somatic cell hybrids have played an important role in human genetics research. They have been especially useful for assigning genes to chromosomes and isolating DNA markers from specific regions of the human genome.^ By employing a combination of somatic cell genetic, recombinant DNA, and cytogenetic techniques, human DNA excision repair gene ERCC4 was mapped regionally to human 16p13.13-13.2, even though the gene has not been cloned. Human x Chinese hamster ovary (CHO) cell hybrids selected for human ERCC4 activity and containing 16p13.1-p13.3 as the only human genetic material were identified. These hybrids were used to order DNA markers located in 16p13.1-p13.3. New DNA markers physically close to ERCC4 were isolated from such hybrids. Using amplified human DNA from the hybrids as probe in fluorescent in situ hybridization, the short arm breakpoint in the chromosome 16 inversion associated with acute myelomonocytic leukemia (AMML) was found to be physically close to the ERCC4 gene. The physical mapping and eventually, the cloning of the ERCC4 gene, will benefit the understanding of the DNA repair system and the study of other important biomedical problems such as tumorigenesis.^ To facilitate the cloning of ERCC4 gene and, in general, the cloning of genes from any defined regions of the human genome, a method was developed for the direct isolation of human transcribed genes ffom somatic cell hybrids. cDNA was prepared from human x rodent hybrid by using consensus 5$\sp\prime$ splice site sequences as primers. These primers were designed to select immature, unspliced messenger RNA (still retaining species specific repeat sequences) as templates. Screening of a derived cDNA library for human repeat sequences resulted in the isolation of human clones at the anticipated frequency with characteristics expected of exons of transcribed human genes. The usefulness of the splice site specific primers was analyzed and the cDNA synthesis conditions with these primers were optimized. The procedure was shown to be sensitive enough to clone weakly expressed genes. Studying the expression of the represented genes with the isolated clones was shown to be feasible. Such regional specific human gene fragments will be very valuable for many human genetic studies such as the search of inherited disease genes and the construction of a cDNA map of the human genome. ^

Functions and intramolecular interactions of ribosomal RNA in decoding of termination codons

Two regions in the 3$\prime$ domain of 16S rRNA (the RNA of the small ribosomal subunit) have been implicated in decoding of termination codons. Using segment-directed PCR random mutagenesis, I isolated 33 translational suppressor mutations in the 3$\prime$ domain of 16S rRNA. Characterization of the mutations by both genetic and biochemical methods indicated that some of the mutations are defective in UGA-specific peptide chain termination and that others may be defective in peptide chain termination at all termination codons. The studies of the mutations at an internal loop in the non-conserved region of helix 44 also indicated that this structure, in a non-conserved region of 16S rRNA, is involved in both peptide chain termination and assembly of 16S rRNA.^ With a suppressible trpA UAG nonsense mutation, a spontaneously arising translational suppressor mutation was isolated in the rrnB operon cloned into a pBR322-derived plasmid. The mutation caused suppression of UAG at two codon positions in trpA but did not suppress UAA or UGA mutations at the same trpA positions. The specificity of the rRNA suppressor mutation suggests that it may cause a defect in UAG-specific peptide chain termination. The mutation is a single nucleotide deletion (G2484$\Delta$) in helix 89 of 23S rRNA (the large RNA of the large ribosomal subunit). The result indicates a functional interaction between two regions of 23S rRNA. Furthermore, it provides suggestive in vivo evidence for the involvement of the peptidyl-transferase center of 23S rRNA in peptide chain termination. The $\Delta$2484 and A1093/$\Delta$2484 (double) mutations were also observed to alter the decoding specificity of the suppressor tRNA lysT(U70), which has a mutation in its acceptor stem. That result suggests that there is an interaction between the stem-loop region of helix 89 of 23S rRNA and the acceptor stem of tRNA during decoding and that the interaction is important for the decoding specificity of tRNA.^ Using gene manipulation procedures, I have constructed a new expression vector to express and purify the cellular protein factors required for a recently developed, realistic in vitro termination assay. The gene for each protein was cloned into the newly constructed vector in such a way that expression yielded a protein with an N-terminal affinity tag, for specific, rapid purification. The amino terminus was engineered so that, after purification, the unwanted N-terminal tag can be completely removed from the protein by thrombin cleavage, yielding a natural amino acid sequence for each protein. I have cloned the genes for EF-G and all three release factors into this new expression vector and the genes for all the other protein factors into a pCAL-n expression vector. These constructs will allow our laboratory group to quickly and inexpensively purify all the protein factors needed for the new in vitro termination assay. (Abstract shortened by UMI.) ^

David Einhorn : Memorial Volume ; seclected sermons and addresses

ed. by Kaufmann Kohler. A biographical essay by Kaufmann Kohler. A memorial oration by Emil G. Hirsch

Purification and cloning of the novel phosphoprotein copine III and characterization of its associated kinase activity

The copines, named and first described by Creutz et al. (1998), comprise a two C2 domain-containing protein family that can aggregate phosphatidylserine membranes in a calcium-dependent manner. Although no enzymatic function has been attributed to copines, their carboxyl terminus shows homology to the A domain found in integrins that allows binding of magnesium ions. The secondary structure of A domains resembles a Rossmann fold, which can bind dinucleotides and is present in a number of intracellular enzymes. Due to a crossreacting activity of Mik b 1, an antibody to the IL-2R b chain, we were able to serendipitously clone human copine III (CIII). CIII is 65% identical to copine I (CI) and the 5 kb CIII transcript is expressed ubiquitously as determined by a multitissue Northern blot. A polyclonal antibody generated against the carboxyl terminus of CIII recognized CIII in immunoblots and immunoprecipitations. Phosphorylation of CIII was observed on serine and threonine residues, as determined by phosphoamino acid analysis. ^ Experiments were designed to determine whether or not any enzymatic activity, specifically kinase activity, was intrinsic to or associated with CIII. In vitro and in gel kinase assays were performed using transfected HA-tagged CI and CIII, immunoprecipitated endogenous CIII and purified endogenous CIII. The exogenous substrate MBP was phosphorylated in all in vitro kinase assays containing CIII protein purification and column chromatography expertise with me. ^

Physical and functional mapping of a novel tumor suppressor locus for prostate cancer within chromosome 10p12.31-q11

Prostate cancer remains the second leading cause of male cancer deaths in the United States, yet the molecular mechanisms underlying this disease remain largely unknown. Cytogenetic and molecular analyses of prostate tumors suggest a consistent association with the loss of chromosome 10. Previously, we have defined a novel tumor suppressor locus PAC-1 within chromosome 10pter-q11. Introduction of the short arm of chromosome 10 into a prostatic adenocarcinoma cell line PC-3H resulted in dramatic tumor suppression and restoration of a programmed cell death pathway. Using a combined approach of comparative genomic hybridization and microsatellite analysis of PC-3H, I have identified a region of hemizygosity within 10p12-p15. This region has been shown to be involved in frequent loss of heterozygosity in gliomas and melanoma. To functionally dissect the region within chromosome 10p containing PAC-1, we developed a strategy of serial microcell fusion, a technique that allows the transfer of defined fragments of chromosome 10p into PC-3H. Serial microcell fusion was used to transfer defined 10p fragments into a mouse A9 fibrosarcoma cell line. Once characterized by FISH and microsatellite analyses, the 10p fragments were subsequently transferred into PC-3H to generate a panel of microcell hybrid clones containing overlapping deletions of chromosome 10p. In vivo and microsatellite analyses of these PC hybrids identified a small chromosome 10p fragment (an estimated 31 Mb in size inclusive of the centromere) that when transferred into the PC-3H background, resulted in significant tumor suppression and limited a region of functional tumor suppressor activity to chromosome 10p12.31-q11. This region coincides with a region of LOH demonstrated in prostate cancer. These studies demonstrate the utility of this approach as a powerful tool to limit regions of functional tumor suppressor activity. Furthermore, these data used in conjunction with data generated by the Human Genome Project lent a focused approach to identify candidate tumor suppressor genes involved in prostate cancer. ^