RNAi as a tool for target discovery in early pharmaceutical research.

The pharmaceutical industry is currently faced with increasing pressure due to patent expirations for block busters, healthcare reforms with strained budgets and growing demands for approval by administrative organizations like the FDA and the EMA. High attrition rates especially in the later expensive stages of the drug development process ask for thoroughly validated drug targets at the beginning of such projects. The great potential of RNA interference strategies toward reaching this goal is outlined in this article.

Shift of syndecan-1 expression from epithelial to stromal cells during progression of solid tumours.

Syndecan-1 (SDC-1), a protein found on cells and in the extracellular matrix, participates in cell proliferation, cell migration and cell-matrix interactions. SDC-1 expression correlates with the maintenance of epithelial morphology and inhibition of invasiveness. In the present study, a second SDC-1 mRNA isoform was identified and the expression of both transcripts was investigated in various normal and malignant tissues. Both transcripts were coexpressed at equal levels in all tissues and organs analysed. Cancer-profiling array (CPA) analysis of 241 non-enriched tumour and normal cDNAs revealed stronger upregulation of SDC-1 in tumour tissues as compared with oligonucleotide array-based expression analysis of SDC-1 in microdissected breast, prostate, lung, and colon carcinoma cells. With in situ hybridisation and immunohistochemistry it was demonstrated that this difference in SDC-1 expression originates from stromal cells present in tumour connective tissue. But only the cells in connective tissue surrounding breast, lung, colon and bladder carcinoma showed upregulation of SDC-1. These stromal cells were characterised as spindle cells with myofibroblastic differentiation and they may contribute to the dedifferentiation of tumour cells and the development of metastasis.

Slow-binding inhibition of Escherichia coli cystathionine beta-lyase by L-aminoethoxyvinylglycine: a kinetic and X-ray study.

The pyridoxal 5'-phosphate (PLP)-dependent cystathionine beta-lyase (CBL) was previously found to be inhibited by the natural toxins rhizobitoxine and l-aminoethoxyvinylglycine (AVG). The present study characterizes the interaction of Escherichia coli CBL with AVG and methoxyvinylglycine (MVG) by a combination of kinetic methods and X-ray crystallography. Upon AVG treatment, time-dependent, slow-binding inhibition [Morrison, J. F. (1982) Trends Biochem. Sci. 7, 102-105] was observed due to the generation of a long-lived, slowly dissociating enzyme-inhibitor complex. Kinetic analysis revealed a one-step inhibition mechanism (CBL + AVG --> CBLAVG, Ki = 1.1 +/- 0.3 microM) with an association rate constant (k1) of 336 +/- 40 M-1 s-1. This value is several orders of magnitude lower than typical bimolecular rate constants of ES formation, suggesting that additional steps occur before formation of the first detectable CBLAVG complex. Loss of activity is paralleled by the conversion of the pyridoxaldimine 426 nm chromophore to a 341 nm-absorbing species. On the basis of the recently solved structure of native CBL [Clausen, T., et al. (1996) J. Mol. Biol. 262, 202-224], it was possible to elucidate the X-ray structure of the CBLAVG complex and to refine it to an R-factor of 16.4% at 2.2 A resolution. The refined structure reveals the geometry of the bound inhibitor and its interactions with residues in the active site of CBL. Both the X-ray structure and the absorbance spectrum of the CBLAVG complex are compatible with a ketimine as the reaction product. Thus, the inhibitor seems to bind in a similar way to CBL as the substrate, but after alpha-proton abstraction, the reaction proceeds in a CBL nontypical manner, i.e. protonation of PLP-C4', resulting in the "dead-end" ketimine PLP derivative. The CBLAVG structure furthermore suggests a binding mode for rhizobitoxine and explains the failure of MVG to inhibit CBL.

Reaction mechanism of Escherichia coli dihydrodipicolinate synthase investigated by X-ray crystallography and NMR spectroscopy.

Dihydrodipicolinate synthase (DHDPS) catalyzes the condensation of pyruvate with L-aspartate beta-semialdehyde. It is the first enzyme unique to the diaminopimelate pathway of lysine biosynthesis. Here we present the crystal structures of five complexes of Escherichia coli DHDPS with substrates, substrate analogs, and inhibitors. These include the complexes of DHDPS with (1) pyruvate, (2) pyruvate and the L-aspartate beta-semialdehyde analog succinate beta-semialdehyde, (3) the inhibitor alpha-ketopimelic acid, (4) dipicolinic acid, and (5) the natural feedback inhibitor L-lysine. The kinetics of inhibition were determined, and the binding site of the L-lysine was identified. NMR experiments were conducted in order to elucidate the nature of the product of the reaction catalyzed by DHDPS. By this method, (4S)-4-hydroxy-2,3,4,5-tetrahydro-(2S)-dipicolinic acid is identified as the only product. A reaction mechanism for DHDPS is proposed, and important features for inhibition are identified.

Crystal structure of the pyridoxal-5'-phosphate dependent cystathionine beta-lyase from Escherichia coli at 1.83 A.

Cystathionine beta-lyase (CBL) is a member of the gamma-family of PLP-dependent enzymes, that cleaves C beta-S bonds of a broad variety of substrates. The crystal structure of CBL from E. coli has been solved using MIR phases in combination with density modification. The structure has been refined to an R-factor of 15.2% at 1.83 A resolution using synchroton radiation diffraction data. The asymmetric unit of the crystal cell (space group C222(1)) contains two monomers related by 2-fold symmetry. A homotetramer with 222 symmetry is built up by crystallographic and non-crystallographic symmetry. Each monomer of CBL can be described in terms of three spatially and functionally different domains. The N-terminal domain (residues 1 to 60) consists of three alpha-helices and one beta-strand. It contributes to tetramer formation and is part of the active site of the adjacent subunit. The second domain (residues 61 to 256) harbors PLP and has an alpha/beta-structure with a seven-stranded beta-sheet as the central part. The remaining C-terminal domain (residues 257 to 395), connected by a long alpha-helix to the PLP-binding domain, consists of four helices packed on the solvent-accessible side of an antiparallel four-stranded beta-sheet. The fold of the C-terminal and the PLP-binding domain and the location of the active site are similar to aminotransferases. Most of the residues in the active site are strongly conserved among the enzymes of the transsulfuration pathway. Additionally, CBL is homologous to the mal gamma gene product indicating an evolutionary relationship between alpha and gamma-family of PLP-dependent enzymes. The structure of the beta, beta, beta-trifluoroalanine inactivated CBL has been refined at 2.3 A resolution to an R-factor of 16.2%. It suggests that Lys210, the PLP-binding residue, mediates the proton transfer between C alpha and S gamma.

Cloning, purification, and crystallization of Escherichia coli cystathionine beta-lyase.

The metC gene coding for cystathionine beta-lyase of Escherichia coli has been cloned and used to construct an overproducing E. coli strain. An efficient purification scheme has been developed and the purified enzyme has been crystallized by the hanging drop vapour diffusion method using either ammonium sulfate or polyethyleneglycol 400 as precipitating agent. The crystals belong to the orthorombic space group C222. Their unit cell parameters are a = 60.9 A, b = 154.7 A and c = 152.7 A. Consideration of the possible values of VM accounts for the presence of one dimer per asymmetric unit. The crystals are suitable for X-ray analysis and a complete native date set to 1.83 A resolution has been collected using synchrotron radiation.

Repression of Acetolactate Synthase Activity through Antisense Inhibition (Molecular and Biochemical Analysis of Transgenic Potato (Solanum tuberosum L. cv Desiree) Plants).

Acetolactate synthase (ALS), the first enzyme in the biosynthetic pathway of leucine, valine, and isoleucine, is the biochemical target of different herbicides. To investigate the effects of repression of ALS activity through antisense gene expression we cloned an ALS gene from potato (Solanum tuberosum L. cv Desiree), constructed a chimeric antisense gene under control of the cauliflower mosaic virus 35S promoter, and created transgenic potato plants through Agrobacterium tumefaciens-mediated gene transfer. Two regenerants revealed severe growth retardation and strong phenotypical effects resembling those caused by ALS-inhibiting herbicides. Antisense gene expression decreased the steady-state level of ALS mRNA in these plants and induced a corresponding decrease in ALS activity of up to 85%. This reduction was sufficient to generate plants almost inviable without amino acid supplementation. In both ALS antisense and herbicide-treated plants, we could exclude accumulation of 2-oxobutyrate and/or 2-aminobutyrate as the reason for the observed deleterious effects, but we detected elevated levels of free amino acids and imbalances in their relative proportions. Thus, antisense inhibition of ALS generated an in vivo model of herbicide action. Furthermore, expression of antisense RNA to the enzyme of interest provides a general method for validation of potential herbicide targets.

Expression of a bacterial gene in transgenic plants confers resistance to the herbicide phenmedipham.

Tobacco plants were genetically engineered to express a detoxifying pathway for the herbicide phenmedipham. A gene from Arthrobacter oxidans strain P52 that encodes an enzyme catalysing the hydrolytic cleavage of the carbamate compound phenmedipham has recently been cloned and sequenced. The coding sequence was fused with a cauliflower mosaic virus 35S promoter and introduced into tobacco plants by Agrobacterium-mediated gene transfer. Transgenic plants expressing high levels of phenmedipham hydrolase exhibited resistance when sprayed with the herbicide at up to ten times the usual field application rate.

Purification and Properties of Cystathionine [gamma]-Synthase from Wheat (Triticum aestivum L.).

Cysthathionine [gamma]-synthase (CS), an enzyme involved in methionine biosynthesis, was purified from an acetone powder prepared from wheat (Triticum aestivum L.). After several chromatographic steps and radiolabeling of the partially purified enzyme with sodium cyanoboro[3H]hydride, a single polypeptide with a molecular weight of 34,500 was isolated by sodium dodecyl sulfate-high performance electrophoresis chromatography. Since the molecular weight of the native enzyme was 155,000, CS apparently consists of four identical subunits. The pyridoxal 5[prime]-phosphate-dependent forward reaction has a pH optimum of 7.5 and follows a hybrid ping-pong mechanism with Km values of 3.6 mM and 0.5 mM for L-homoserine phosphate and L-cysteine, respectively. L-Cysteine methyl ester, thioglycolate methyl ester, and sodium sulfide were also utilized as thiol substrates. The latter observation suggests that CS and phosphohomoserine sulfhydrase might be a single enzyme. CS does not seem to be a regulatory enzyme but was irreversibly inhibited by DL-propargylglycine (Ki = 45 [mu]M, Kinact = 0.16 min-1). Furthermore, the homoserine phosphate analogs 4-(phosphonomethyl)-pyridine-2-carboxylic acid, Z-3-(2-phosphonoethen-1-yl)pyridine-2-carboxylic acid, and DL-E-2-amino-5-phosphono-3-pentenoic acid acted as reversible competitive inhibitors with Ki values of 45, 40, and 1.1 [mu]M, respectively.

Mechanisms of interaction of Escherichia coli threonine synthase with substrates and inhibitors.

Threonine synthase (TS), the last enzyme of the threonine biosynthetic pathway, catalyzes L-threonine formation from L-homoserine phosphate (HSerP; Km = 0.5 mM, V = 440 min-1) and DL-vinylglycine. Furthermore, TS catalyzes beta-elimination reactions with L-serine (Km = 150 mM, V = 4.7 min-1), DL-3-chloroalanine, L-threonine, and L-allo-threonine as substrates to yield pyruvate or alpha-ketobutyrate, while L-alanine, L-2-aminobutanoic acid, and L-2-amino-5-phosphonopentanoic acid are substrates for half-transamination reactions to form the pyridoxamine form of the enzyme and the corresponding alpha-keto acid. Spectral analyses of all these reactions revealed the transient formation of strongly absorbing long-wavelength chromophores (lambda max = 440-445 nm), implying the accumulation of the corresponding pyridoxaldimine p-quinonoidal intermediates. HSerP turnover was competitively inhibited by L-3-hydroxyhomoserine phosphate 1 (Ki = 0.050 mM), L-2,3-methanohomoserine phosphate 2 (Ki = 0.010 mM), L-2-amino-3-[(phosphonomethyl)thio)]propanoic acid 5 (Ki = 0.011 mM) and DL-E-2-amino-5-phosphono-4-pentenoic acid 10 (Ki = 0.54 mM). 5 and 10 induced the formation of long-wavelength quinonoidal chromophores (lambda max = 458 and 460 mm, epsilon 47,000 and 30,000 M-1 cm-1), while incubation with either 1 or 2 induced only minor spectral changes. DL-2-Amino-3-[(phosphonomethyl)amino)]propanoic acid inactivated TS (Ki = 0.057 mM, kinact = 1.44 min-1) with 1:1 stoichiometry, transient formation of a 450-nm chromophore, and finally bleaching of any absorbance at wavelengths longer than 320 nm. Z-2-Amino-5-phosphono-3-pentenoic acid 8 is the unusual amino acid found in the peptide antibiotics of the plumbemicin and rhizocticin families. Racemic 8 irreversibly inhibited TS (Ki = 0.1 mM, kinact = 1.50 min-1) with 1:1 stoichiometry and the concomitant formation of a 482-nm chromophore (epsilon approximately 30,000 M-1 cm-1). DL-E-2-Amino-5-phosphono-3-pentenoic acid was a less potent irreversible inhibitor of TS (Ki = 0.4 mM, kinact = 0.25 min-1), inducing absorption maxima at 462 and 500 nm. The acetylenic amino acid DL-2-amino-5-phosphono-4-pentynoic acid 12 bound to TS (KD = 0.38 mM) forming a quinonoidal chromophore (lambda max = 452 nm, epsilon approximately 30,000 M-1 cm-1), but inhibition of the enzyme by 12 could not be detected under assay conditions even at high inhibitor concentrations. Mechanisms consistent with these observations are proposed.

A visible marker for antisense mRNA expression in plants: inhibition of chlorophyll synthesis with a glutamate-1-semialdehyde aminotransferase antisense gene.

Glutamate 1-semialdehyde aminotransferase [(S)-4-amino-5-oxopentanoate 4,5-aminomutase, EC 5.4.3.8] catalyzes the last step in the conversion of glutamate to delta-aminolevulinate of which eight molecules are needed to synthesize a chlorophyll molecule. Two full-length cDNA clones that probably represent the homeologous Gsa genes of the two tobacco (Nicotiana tabacum) genomes have been isolated. The deduced amino acid sequences of the 468-residue-long precursor polypeptides differ by 10 amino acids. The cDNA sequence of isoenzyme 2 was inserted in reverse orientation under the control of a cauliflower mosaic virus 35S promoter derivative in an expression vector and was introduced by Agrobacterium-mediated transformation into tobacco plants. Antisense gene expression decreased the steady-state mRNA level of glutamate 1-semialdehyde aminotransferase, the translation of the enzyme, and chlorophyll synthesis. Remarkably, partial or complete suppression of the aminotransferase mimics in tobacco a wide variety of chlorophyll variegation patterns caused by nuclear or organelle gene mutations in different higher plants. The antisense gene is inherited as a dominant marker.

Inactivation of Escherichia coli threonine synthase by DL-Z-2-amino-5-phosphono-3-pentenoic acid.

The rhizocticines and plumbemicines are two groups of di- and tripeptid antibiotics thought to interfere with threonine or threonine-related metabolism. Z-2-amino-5-phosphono-3-pentenoic acid, the common unusual amino acid constituent of the rhizocticines and plumbemicines, was found to irreversibly inhibit Escherichia coli threonine synthase in a time-dependent reaction that followed pseudo-first order and saturation kinetics. These data provide evidence that the toxicity of the rhizocticines and plumbemicines is due to the inhibition of threonine synthase by Z-2-amino-5-phosphone-3-pentenoic acid, which is liberated by peptidases after uptake into the target cell. Additionally, methods for the purification of threonine synthase from an overproducing E. coli strain and for the enzymatic synthesis of L-homoserine phosphate are described.

Purification and properties of an Arthrobacter oxydans P52 carbamate hydrolase specific for the herbicide phenmedipham and nucleotide sequence of the corresponding gene.

Arthrobacter oxydans P52 isolated from soil samples was found to degrade the phenylcarbamate herbicides phenmedipham and desmedipham cometabolically by hydrolyzing their central carbamate linkages. The phenylcarbamate hydrolase (phenmedipham hydrolase) responsible for the degradative reaction was purified to homogeneity. The enzyme was shown to be a monomer with a molecular weight of 55,000. A 41-kb wild-type plasmid (pHP52) was identified in A. oxydans P52, but not in a derivative of this strain that had spontaneously lost the ability to hydrolyze phenylcarbamates, indicating that the gene for phenylcarbamate degradation (pcd) is plasmid encoded. Determination of two partial amino acid sequences allowed the localization of the coding sequence of the pcd gene on a 3.3-kb PstI restriction fragment within pHP52 DNA by hybridization with synthetic oligonucleotides. The phenylcarbamate hydrolase was functionally expressed in Escherichia coli under control of the lacZ promoter after the 3.3-kb PstI fragment was subcloned into the vector pUC19. A stretch of 1,864 bases within the cloned Pst fragment was sequenced. Sequence analysis revealed an open reading frame of 1,479 bases containing the amino acid partial sequences determined for the purified enzyme. Sequence comparisons revealed significant homology between the pcd gene product and the amino acid sequences of esterases of eukaryotic origin. Subsequently, it was demonstrated that the esterase substrate p-nitrophenylbutyrate is hydrolyzed by phenmedipham hydrolase.

Yeast homoserine kinase. Characteristics of the corresponding gene, THR1, and the purified enzyme, and evolutionary relationships with other enzymes of threonine metabolism.

THR1, the gene from Saccharomyces cerevisiae, encoding homoserine kinase, one of the threonine biosynthetic enzymes, has been cloned by complementation. The nucleotide sequence of a 3.1-kb region carrying this gene reveals an open reading frame of 356 codons, corresponding to about 40 kDa for the encoded protein. The presence of three canonical GCN4 regulatory sequences in the upstream flanking region suggests that the expression of THR1 is under the general amino acid control. In parallel, the enzyme was purified by four consecutive column chromatographies, monitoring homoserine kinase activity. In SDS gel electrophoresis, homoserine kinase migrates like a 40-kDa protein; the native enzyme appears to be a homodimer. The sequence of the first 15 NH2-terminal amino acids, as determined by automated Edman degradation, is in accordance with the amino acid sequence deduced from the nucleotide sequence. Computer-assisted comparison of the yeast enzyme with the corresponding activities from bacterial sources showed that several segments among these proteins are highly conserved. Furthermore, the observed homology patterns suggest that the ancestral sequences might have been composed from separate (functional) domains. A block of very similar amino acids is found in the homoserine kinases towards the carboxy terminus that is also present in many other proteins involved in threonine (or serine) metabolism; this motif, therefore, may represent the binding site for the hydroxyamino acids. Limited similarity was detected between a motif conserved among the homoserine kinases and consensus sequences found in other mono- or dinucleotide-binding proteins.

The human VK locus. Characterization of a duplicated region encoding 28 different immunoglobulin genes.

Two large regions of the human multigene family coding for the variable parts of the immunoglobulin light chains of the K type (VK) have been characterized on cosmid clones. The two germline regions, called Aa and Ab, span together 250,000 base-pairs and comprise 28 different VK gene segments, nine of which have been sequenced. There is a preponderance of VKII genes but genes belonging to subgroups I and III, and genes that cannot be easily assigned to one of the known subgroups, are interspersed within the VKII gene clusters. A number of pseudogenes have been identified. Within the Aa and Ab regions, all gene segments are organized in the same transcriptional orientation. The regions Aa and Ab, whose restriction maps are highly homologous, were shown not to be allelic structures; they must have arisen by a duplication event. Taken together with previous results, one can conclude that the major part of the VK locus exists in duplicated form. One individual has been found who has only one copy of some of the duplicated regions. By chromosomal walking, the A regions could be linked to the O regions, an analysis of which has been reported. The A regions contribute about one-third of the VK genes so far identified.

The human V kappa locus. Characterization of extended immunoglobulin gene regions by cosmid cloning.

As part of the ongoing work in our laboratory on the structural organization of the human V kappa locus we screened cosmid libraries with V kappa gene probes and obtained numerous V kappa gene-containing cosmid clones. Several genomic regions of the V kappa locus were reconstructed from overlapping cosmid inserts and were extended by one step of chromosomal walking. The regions that are called Wa, Wb, Oa, Ob and Ob' comprise about 370 kb (10(3) bases) of DNA and contain 24 V kappa genes and pseudogenes. The V kappa genes belong to the three dominant subgroups (V kappa I, V kappa II, V kappa III) and are arranged to form mixed clusters with members of the different subgroups being intermingled with each other. The distances between the genes range from 1 to 15 kb. Three genes of the Wa and Wb regions that were sequenced turned out to be pseudogenes. Terminal parts of the regions Wa and Ob that do not contain V kappa genes of one of the known subgroups may represent extended spacer regions within the V kappa locus. Wa and Wb are duplicated regions located at different positions of the locus. Region Wb was found to comprise inversely repeated sections of at least 14 kb each that contain V kappa genes oriented in opposite polarity. This finding is consistent with inversion-deletion models of V-J joining; it also shows that the V kappa locus contains not only unique and duplicated but also triplicated parts. The data on the W and O regions are discussed together with those on the L regions and on other regions established in our laboratory. Although the picture of the human V kappa locus with, to date, about 70 different non-allelic V kappa genes is still incomplete, some general features with respect to the organization of the genes and the limited duplication of genomic regions have emerged.

Dispersed human immunoglobulin kappa light-chain genes.

The gene segments encoding the constant and variable regions of human immunoglobulin light chains of the kappa type (C kappa, V kappa) have been localized to chromosome 2. The distance between the C kappa and V kappa genes and the number of germline V kappa genes are unknown. As part of our work on the human V kappa locus, we have now mapped two solitary V kappa gene and a cluster of three V kappa genes to chromosomes 1, 15 and 22, respectively. The three genes that have been sequenced are nonprocessed pseudogenes, and the same may be true for the other two genes. This is the first time that V-gene segments have been found outside the C-gene-containing chromosomes. Our finding is relevant to current estimates of the size of the V kappa-gene repertoire. Furthermore, the dispersed gene regions have some unusual characteristics which may help to clarify the mechanism of dispersion.

Subgroup IV of human immunoglobulin K light chains is encoded by a single germline gene.

The series of studies on the human K light chain genes of the various subgroups is concluded by this report on the isolation and nucleotide sequence determination of a functional VKIV gene (abbreviations ref. 1) and its germline counterpart. The rearranged gene which stems from a lymphoid cell line and the germline gene differ in four nucleotides which can be attributed to somatic mutations; three of the mutations are clustered in CDR3. The germline gene regions of two unrelated individuals were identical over a stretch of 1267 bp. By hybridization experiments it is shown that the human K locus contains only one VKIV gene. In 16 lymphoid cell lines studied here, the VKIV gene is frequently deleted or aberrantly rearranged which may be a consequence of peculiarities of its function and/or its structural organization.

A large section of the gene locus encoding human immunoglobulin variable regions of the kappa type is duplicated.

The structure of a new segment of the gene locus encoding the variable regions of human immunoglobulins of the Kappa type (VK) has been elucidated. This segment (cluster B) encompasses six VK sequences, which belong to three different subgroups and which are arranged in the same transcriptional orientation. Part of cluster B was found to be very similar to another region of the VK gene locus, which was cloned previously (cluster A). Sequence differences between the homologous region of clusters A and B range from 0.2% to 3.7% depending on the position of the VK sequences. The divergence is in the same range for genes and pseudogenes. Hybridization experiments with DNAs from different individuals clearly demonstrate that the two segments are located at different positions within the VK locus and do not represent allelic variants. The sequence homology between clusters A and B is higher than the homology of both clusters to an allelic variant, which is represented by a DNA segment that had been isolated from another individual. These results, together with a report in the literature of two other homologous regions in the VK locus, make it very likely that a major part of even the whole locus is duplicated. In this case, VK gene numbers would be higher than previously estimated on the basis of hybridization studies. An inverse orientation of VK gene clusters would explain published data on rearrangement products in B-cells if an inversion-deletion mechanism is assumed.