Primary structure of rat hepatocyte growth factor and induction of its mRNA during liver regeneration following hepatic injury

Overlapping cDNA clones for rat hepatocyte growth factor (rHGF) were isolated by cross-hybridization with the cloned cDNA for human hepatocyte growth factor (hHGF) and the nucleotide sequence of the cDNA was determined. The entire primary structure of rHGF was deduced from the sequence. Comparison of the amino acid sequences between rat and human HGFs revealed that the two sequences are highly conserved throughout the protein structures, suggesting that rat and human HGFs may be functionally similar. Responses of the rHGF mRNA during liver regeneration in rats were examined by Northern blot hybridization analysis with the aid of the cDNA probe for rHGF. The mRNA levels increased in the liver and spleen but not in the kidney after administration of carbon tetrachloride. At the maximum level of induction, the rHGF mRNA increased in the liver about 4.5-fold over its normal level. The mRNA levels also increased in the liver and spleen after administration of D-galactosamine. On the other hand, no obvious increase of the mRNA was observed in the liver and spleen after partial hepatectomy. These observations suggest that HGF may function as a regulator of liver regeneration following hepatic injury caused by hepatotoxins.     

Liver (B-type) phosphofructokinase mRNA. Cloning, structure, and expression

Mouse liver mRNA enriched in sequences coding for liver phosphofructokinase by polysome immunoadsorption was used as a template for the synthesis of cDNA. The double-stranded cDNA was inserted into the expression vector lambda gt11 and cloned. Preliminary identification of clones containing cDNA sequences for phosphofructokinase was made by screening the library with anti-rat liver phosphofructokinase serum and horseradish peroxidase-conjugated goat anti-rabbit IgG as second antibody. Subsequently, by selecting antibodies specific to fusion proteins expressed by putative clones and by reacting with Western blots of mouse liver proteins several clones were positively identified as containing liver phosphofructokinase sequences. A cDNA clone corresponding to 2708 nucleotides of liver phosphofructokinase mRNA was further characterized and sequenced. The liver phosphofructokinase mRNA has an open reading frame of 2343 nucleotides followed by a 3'-untranslated region of 303 nucleotides. The G/C-rich (76%) portion of the 5'-untranslated region precedes a characteristic translational start site of CCGCC(AUG). The mRNA coding sequence indicates that the liver phosphofructokinase subunit is composed of 780 amino acid residues and has a Mr of 85,000. Comparison of the deduced amino acid sequence of mouse liver phosphofructokinase with the known rabbit muscle phosphofructokinase shows 68% homology. The N-half of the liver phosphofructokinase has conserved substrate binding sites for ATP and fructose-6-P. The 25 C-terminal residues, which contain the ATP inhibitory site, are the least homologous (20%) but contain a putative phosphorylation site (Arg-Arg-X-X-Ser). The liver phosphofructokinase mRNA is under nutritional and hormonal regulation. The liver phosphofructokinase mRNA level increased 4-fold when previously starved mice were refed a high carbohydrate, fat-free diet. This increase in mRNA level was blocked by 50% by the administration of dibutyryl cAMP. The induction of liver phosphofructokinase mRNA by fasting/refeeding was also diminished in streptozotocin diabetic mice.     

Regulation of cholesterol 7 alpha-hydroxylase in the liver. Cloning, sequencing, and regulation of cholesterol 7 alpha-hydroxylase mRNA.

Monospecific antibody against purified rat liver cholesterol 7 alpha-hydroxylase cytochrome P-450 was used to screen a lambda gt11 cDNA library constructed from immuno-enriched polysomal RNA of cholestyramine-treated female rat liver. Two types of cDNA clones differing in the length of the 3'-untranslated region were identified, and DNA sequences were determined. The full length clone contains 3561 base pairs plus a long poly(A) tail. The amino acid sequence deduced from the open reading frame revealed a unique P-450 protein containing 503 amino acid residues which belonged to a new gene family designated family VII or CYP7. Southern blot hybridization experiments indicated that the minimal size of P-450 VII gene was 11 kilobase pairs (kb), and there was probably only one gene in this new family. Northern blot hybridization using specific cDNA probes revealed at least two major mRNA species of about 4.0 kb and 2.1 kb, respectively. These two mRNA species may be derived from the use of different polyadenylation signals and reverse-transcribed to two types of cDNA clones. Cholesterol 7 alpha-hydroxylase mRNAs were induced 2- to 3-fold in rat liver by cholestyramine treatment. The mRNA level was rapidly reduced upon the removal of the inducer. Similarly, cholesterol feeding induced enzyme activity, protein, and mRNA levels in the rat by 2-fold, suggesting that cholesterol is an important regulator of cholesterol 7 alpha-hydroxylase in the liver. On the other hand, dexamethasone and pregnenolone-16 alpha-carbonitrile drastically reduced the activity, protein, and mRNA levels. These experiments suggest that the induction of cholesterol 7 alpha-hydroxylase activity by cholestyramine or cholesterol and inhibition of cholesterol 7 alpha-hydroxylase activity by bile acid feedback are results of the rapid turnover of cholesterol 7 alpha-hydroxylase enzyme and mRNA levels.     

Fatty acid delta(5)-desaturase mRNA is regulated by dietary vitamin A and exogenous retinoic acid in liver of adult rats.

Delta(5)-Desaturase (D5D) catalyzes the Delta(5,6) desaturation of dietary essential fatty acids of the n-6 and n-3 series. By subtraction hybridization of vitamin A (VA)-deficient and control rat liver cDNA libraries, we isolated a 106-bp cDNA fragment that proved to be homologous to human liver D5D cDNA and used it as a probe to analyze rat D5D mRNA and clone the rat full-length cDNA. Delta(5)-Desaturase mRNA was threefold more abundant in liver from VA-deficient rats than in liver from VA-sufficient rats and was expressed dose dependently when dietary VA was varied (VA marginal > control > VA supplemented). Treatment of VA-deficient rats with all-trans-retinoic acid lowered the level of expression of D5D mRNA toward that of VA-sufficient rats. The 3413-bp full-length D5D cDNA cloned from rat liver contains an open reading frame of 447 amino acid residues sharing 92% similarity with its human counterpart. Expression of this cDNA in HEK293T cells incubated with dihomo-gamma-linolenic acid (20:3, n-6) resulted in a significantly increased ratio of the product, arachidonic acid (20:4, n-6), to substrate in cell lipid extracts. Delta(5)-Desaturase mRNA is expressed in relatively high abundance in rat adrenal gland and mammary tissue and moderately in liver, kidney, lung, spleen, thymus, brain, and eye. The regulation of D5D by VA could be important for growth and development, and reproduction, as well as in the control of inflammation.     

Rat corticosteroid binding globulin: primary structure and messenger ribonucleic acid levels in the liver under different physiological conditions

Rat corticosteroid binding globulin (CBG) cDNAs were isolated from a lambda gt11 liver cDNA library. When rat hepatic mRNA was hybrid selected and translated in vitro, a major product reacted with antibodies against rat CBG and its Mr (approximately 43,000) was consistent with a nonglycosylated, CBG precursor polypeptide. Two overlapping cDNAs produced a 1,432 nucleotide sequence with an open reading frame comprising 396 amino acids. This includes a potential signal peptide of 22 residues followed by the amino terminus of purified rat CBG. Rat CBG therefore contains 374 amino acids (Mr = 42,196), and has six consensus sites for N-glycosylation. There is 60% identity in the primary structures of rat and human CBG over 383 residues that comprise the human sequence. Furthermore, the single cysteine in rat CBG corresponds to one of two cysteines in human CBG, and this may be significant because a cysteine is located in the human CBG steroid binding site. Northern analysis of RNA from various rat tissues revealed an approximate 1.8 kilobase CBG mRNA only in the liver. Its relative abundance in a pregnant rat was only 30% higher than in an adult female; approximately 3-fold higher than in an adult male, and 25-fold higher than in the fetuses from the same animal. Southern analysis of rat genomic DNA suggests the presence of a single gene for CBG.     

Molecular cloning of a cDNA for human pyruvate carboxylase. Structural relationship to other biotin-containing carboxylases and regulation of mRNA content in differentiating preadipocytes

An oligonucleotide probe specific for the amino acid sequence at the biotin site in pyruvate carboxylase was used to screen a human liver cDNA library. Nine cDNA clones were isolated and three proved to be pyruvate carboxylase clones based on nucleotide sequencing and Northern blotting. The biotin site amino acid sequence of human pyruvate carboxylase agreed perfectly with that of the sheep enzyme in 14 consecutive positions. The highly conserved amino acid sequence, Ala-Met-Lys-Met, found at the biotin site in most biotin-containing carboxylases was also present in human pyruvate carboxylase. The termination codon was located 35 residues 3' to the lysine residue at which the biotin is attached. Therefore, the biotin cofactor is covalently linked near the carboxyl-terminal end of the carboxylase protein. These data are consistent with that observed for other biotin-containing carboxylases and strongly suggests that the genes encoding the biotin-containing carboxylases may have evolved from a common ancestral gene. Northern blotting of mRNA isolated from human, baboon, and rat liver demonstrated that the pyruvate carboxylase mRNA was 4.2 kilobase pairs in length in all species examined. Southern blot analysis of genomic DNA isolated from human-Chinese hamster somatic cell hybrids localized the pyruvate carboxylase gene on the long arm of human chromosome 11. The human cDNA was also used to quantitate pyruvate carboxylase mRNA levels in a differentiating mouse preadipocyte cell line. These data demonstrated that pyruvate carboxylase mRNA content increased 23-fold in 7 days after the onset of differentiation.     

Differential expression of the human thymosin-beta 4 gene in lymphocytes, macrophages, and granulocytes

A cDNA clone encoding human thymosin-beta 4 was isolated from a cDNA library prepared from peripheral blood leukocytes of a patient with acute lymphocytic leukemia. This clone contained the entire coding sequence of 43 amino acid residues of thymosin-beta 4 and had an initiation codon and two termination codons. The amino acid and nucleotide sequences in the coding region were well conserved between rat and human. Nine of 132 nucleotides were different in the coding sequences (93% homology), but the deduced amino acid sequences were identical. No signal peptide was found in the deduced protein sequence. Human thymosin-beta 4 mRNA, approximately 830 nucleotides in length, was about 30 nucleotides larger than rat thymosin-beta 4 mRNA. Expression of the human thymosin-beta 4 gene in various primary myeloid and lymphoid malignant cells and in a few human hemopoietic cell lines was studied. Northern blot analyses of different neoplastic B lymphocytes revealed that steady state levels of thymosin-beta 4 mRNA varied as a function of differentiation stage. Thymosin-beta 4 mRNA levels were decreased in myeloma cells as are class II human leukocyte antigen, Fc receptor, and complement receptor, suggesting a relationship between thymosin-beta 4 and the immune response. Thymosin-beta 4 mRNA was more highly expressed in mature granulocytes than in immature blastic cells. Treatment of THP-1 cells, a human monocytic cell line, with recombinant human interferon-lambda reduced the levels of thymosin-beta 4 mRNA. Its level decreased after differentiation of THP-1 cells into Ia+ macrophages, but increased after differentiation of HL-60 cells into Ia- macrophages. The pattern of thymosin-beta 4 gene expression suggests that it may play a fundamental role in the host defense mechanism.     

cDNA cloning, functional expression and cellular localization of rat liver mitochondrial electron-transfer flavoprotein-ubiquinone oxidoreductase protein

A 64 kD protein was enriched from rat liver mito-chondria during the purification of choline dehydro-genase (CHDH)[1]. Homologous comparison and func-tional experiments demonstrated that the protein was electron-transfer flavoprotein-ubiquinone oxidoreduc-tase protein (ETF-QO). The N-terminal sequence determination of rat liver ETF-QO protein purified by various methods did not provide unequivocal result. However, when the protein was digested with V8 protease, peptide fragments could b…     

Structure and expression of dog apolipoprotein A-I, E, and C-I mRNAs: implications for the evolution and functional constraints of apolipoprotein structure

Dog apolipoprotein (apo) C-I, A-I, and E cDNA clones were identified in a dog liver cDNA library in lambda gt10 by hybridization to synthetic oligonucleotide probes with the corresponding human DNA sequences. The longest clone for each apolipoprotein was completely sequenced. The apoC-I cDNA sequence predicts a protein of 62 residue mature peptide preceded by a 26 amino acid signal peptide. The apoA-I cDNA sequence predicts a 242 residue mature peptide, a 6 residue pro-segment, and an 18 residue signal peptide. The apoE cDNA, which lacks the signal peptide region, predicts a mature peptide of 291 amino acid residues. Slot blot hybridization of total RNA isolated from various dog tissues to dog apoC-I, A-I, and E cDNA probes indicates that apoC-I mRNA is detectable in liver only, apoA-I mRNA is present in liver and small intestine, though the concentration in the latter tissue is only approximately 15% of that in the liver, and apoE mRNA is present in multiple tissues including liver, jejunum, urinary bladder, ileum, colon, brain, kidney, spleen, pancreas, and testis with relative concentrations (%) of 100, 17.5, 7.5, 6.9, 5.9, 5.5, 5.0, 3.3, 1.0, and 1.0, respectively. These tissue distributions indicate that nascent lipoprotein particles produced in the dog small intestine would contain apoA-I and apoE but not apoC-I. The widespread tissue distribution of apoE mRNA indicates that like other mammals, peripheral synthesis of apoE contributes significantly to the total apoE pool in dog. We next compared the cDNA sequences among different vertebrate species for apoC-I (human and dog), A-I (human, rat, dog, rabbit and chicken), and E (human, rat, dog and rabbit) and calculated the rate of nucleotide substitution for each gene. Our results indicate that apoC-I has evolved rather rapidly and that on the whole, apoA-I is more conservative than apoE, contradictory to an earlier suggestion. ApoA-I is also more conservative than a region (residues 4204-4536) at the carboxyl-terminal portion, but less conservative than a region (residues 595-979) at the amino-terminal portion of apoB-100. Some regions in each of the apolipoproteins studied are better conserved than others and the rate of evolution of individual regions seems to be related to the stringency of functional requirements. Finally, we estimate that the human apoC-I pseudogene arose more than 35 million years ago, becoming nonfunctional soon after its formation.     

Human hepatic lipase. Cloned cDNA sequence, restriction fragment length polymorphisms, chromosomal localization, and evolutionary relationships with lipoprotein lipase and pancreatic lipase

Human hepatic lipase is an important enzyme in high density lipoprotein (HDL) metabolism, being implicated in the conversion of HDL2 to HDL3. Three human hepatic lipase cDNA clones were identified in two lambda gt11 libraries from human liver. The cDNA-derived amino acid sequence predicts a protein of 476 amino acid residues, preceded by a 23-residue signal peptide. Four potential N-glycosylation sites are identified, two of which are conserved in rat hepatic lipase. On alignment with human, mouse, and bovine lipoprotein lipase, the same two sites were also conserved in lipoprotein lipase in all three species. Stringent conservation of the cysteine residues was also evident. Comparative analysis of amino acid sequences shows that hepatic lipase evolves at a rapid rate, 2.07 x 10(-9) substitutions/site/year, about four times that in lipoprotein lipase and half that in pancreatic lipase. Further, hepatic lipase and pancreatic lipase appear to be evolutionarily closer to each other than either of them is to lipoprotein lipase. Southern blot analysis revealed high frequency restriction fragment length polymorphisms of the hepatic lipase gene for the enzymes HindIII and MspI. these polymorphisms will be useful for haplotype and linkage analysis of the hepatic lipase gene. Using cloned human hepatic lipase cDNA as a hybridization probe, we performed Southern blot analysis of a panel of 13 human-rodent somatic cell hybrids. Concordance analysis of the various hybrid clones indicates that the hepatic lipase gene is located on the long arm of human chromosome 15. Analysis of hybrids containing different translocations of chromosome 15 localized the gene to the region 15q15----q22.     

牛肝辅酶Ⅱ依赖性视黄醇脱氢酶cDNA的克隆及组织表达

迄今为止的研究证明 ,维生素A亦称视黄醇(retinol)的生理功能是通过其两步氧化代谢产物视黄醛与视黄酸 (亦称维甲酸 )来完成的 .视黄醛通过其光学异构体 1 1 顺式视黄醛与视觉细胞内的视蛋白 (opsin)结合组成视色素 .感光时 ,1 1 顺式视黄醛转变成全反式视黄醛从视蛋白脱落 ,这一过程同时传导到大脑产生视觉[1 ] .全反式维甲酸 (all transretinoicacid)则通过与其在核内受体 (RARα ,β ,γ)结合调节基因的转录来发挥其许多重要的生理功能 ,包括正常胚胎的发育 ,形态、神经系统的形成 ,成体动物的生长、发育、繁殖等 ,并通过调解组织及…     

cDNA cloning,functional expression and cellular localization of rat liver mitochondrial electron-transfer flavoprotein-ubiquinone oxidoreductase protein

A membrane-bound protein was purified from rat liver mitochondria. After being digested with V8 protease, two peptides containing identical 14 amino acid residue sequences were obtained. Using the 14 amino acid peptide derived DNA sequence as gene specific primer, the cDNA of correspondent gene 5′-terminal and 3′-terminal were obtained by RACE technique. The full-length cDNAthat encoded a protein of 616 amino acids was thus cloned, which included the above mentioned peptide sequence. The full length cDNA was highly homologous to that of human ETF-QO, indicating that it may be the cDNA of rat ETF-QO. ETF-QO is an iron sulfur protein located in mitochondria inner membrane containing two kinds of redox center: FAD and [4Fe-4S] center. After comparing the sequence from the cDNA of the 616 amino acids protein with that of the mature protein of rat liver mitochondria, it was found that the N terminal 32 amino acid residues did not exist in the mature protein, indicating that the cDNA was that of ETF-QOp. When the cDNA was expressed in Saccharomyces cerevisiae with inducible vectors, the protein product was enriched in mitochondrial fraction and exhibited electron transfer activity (NBT reductase activity) of ETF-QO. Results demonstrated that the 32 amino acid peptide was a mitochondrial targeting peptide, and both FAD and iron-sulfur cluster were inserted properly into the expressed ETF-QO. ETF-QO had a high level expression in rat heart, liver and kidney. The fusion protein of GFP-ETF-QO co-localized with mitochondria in COS-7 cells.     

Molecular cloning of the full-length cDNA encoding mouse neutral ceramidase. A novel but highly conserved gene family of neutral/alkaline ceramidases

We report here the molecular cloning, sequencing, and expression of the gene encoding the mouse neutral ceramidase, which has been proposed to function in sphingolipid signaling. A full-length cDNA encoding the neutral ceramidase was cloned from a cDNA library of mouse liver using the partial amino acid sequences of the purified mouse liver ceramidase. The open reading frame of 2,268 nucleotides encoded a polypeptide of 756 amino acids having nine putative N-glycosylation sites. Northern blot analysis revealed that the mRNA of the ceramidase was expressed widely in mouse tissues, with especially strong signals found in the liver and kidney. The ceramidase activity of lysates of CHOP cells increased more than 900-fold when the cells were transformed with a plasmid containing the cDNA encoding ceramidase. We also cloned the ceramidase homologue from the cDNA library of mouse brain and found that the sequence of the open reading frame, but not the 5'-noncoding region, was identical to that of the liver. Interestingly, phylogenetic analysis of various ceramidases clearly indicated that neutral/alkaline ceramidases form a novel but highly conserved gene family that is evolutionarily different from lysosomal acid ceramidases.     

Rat apolipoprotein C-II lacks the conserved site for proteolytic cleavage of the pro-form.

Apolipoprotein C-II (apoC-II) plays a critical role in the metabolism of plasma lipoproteins as an activator for lipoprotein lipase. Human apoC-II consists of 79 amino acid residues (pro-apoC-II). A minor fraction is converted to a mature form by cleavage at the site QQDE releasing the 6 amino-terminal residues. We have cloned and sequenced the cDNA for rat apoC-II from a liver cDNA library using human apoC-II cDNA as a probe. The cDNA encodes a protein of 97 amino acid residues including a signal peptide of 22 amino acid residues. There is approximately 60% similarity between the deduced amino acid sequence of rat apoC-II and other apoC-II sequences presently known (human, monkey, dog, cow, and guinea pig). Compared to these, rat apoC-II is one residue shorter at the carboxyl terminus. Furthermore, there is a deletion of 3 amino acid residues (PQQ) in the highly conserved cleavage site where processing from pro- to mature apoC-II occurs in other species. Accordingly, rat apoC-II isolated from plasma was mainly in the pro-form. Northern blot analyses indicated that rat apoC-II is expressed both in liver and in small intestine.     

Alternative splicing of glucokinase mRNA in rat liver.

The sequences of two near full-length cDNAs encoding rat liver glucokinase are reported. One of the cDNAs is essentially identical to the cDNA cloned by Andreone, Printz, Pilkis, Magnuson & Granner. [(1989) J. Biol. Chem. 264, 363-369]. The other cDNA contains a 151 bp insertion and a downstream 52 bp deletion. The inserted block of bases has been shown to originate from an optional cassette exon, termed 2A, between the previously described exons 1 and 2. The conceptual translation product from the variant mRNA is identical to the original glucokinase protein for the first 15 amino acids. Next there is a novel polypeptide sequence of 87 residues, comprising 50 residues encoded by the cassette exon and 37 residues specified by an altered reading frame in exon 2. Due to the 52 bp deletion, 17 amino acids of the reference sequence are then missing, after which the sequence reverts to the original. Northern blot analysis with oligonucleotide probes has shown that alternatively spliced mRNA represents about 5% of total glucokinase mRNA. Alternative splicing of glucokinase mRNA in liver may explain earlier findings of minor isoforms of hepatic glucokinase.     

Sequence and acute phase regulation of rat alpha 1-inhibitor III messenger RNA

cDNA clones coding for the plasma proteinase inhibitor alpha 1-inhibitor III were isolated from an acute phase rat liver library. The isolates could be divided into four groups with characteristic BamHI restriction fragment patterns. The identity of the prototype clone pRLA1I3/2J was established by comparison with the published amino acid sequence of the purified protein. It codes for a 1477-amino acid precursor polypeptide with a 24-residue signal peptide. The mature protein shares 58% overall sequence identity with rat alpha 2-macroglobulin and contains a typical internal thiolester sequence. Twenty-two of its twenty-three cysteinyl residues are conserved with alpha 2-macroglobulin implying similar tertiary structure. However, the prototype alpha 1-inhibitor III sequence differed significantly from the rat and human alpha 2-macroglobulin sequences in its bait region suggesting alpha 1-inhibitor III possesses proteinase inhibitory specificities different from those of alpha 2-macroglobulin. The variant alpha 1-inhibitor III clone pRLA1I3/2J from a second cDNA group also differed from the prototype in the bait region coding sequence, although both specify similar signal peptides and NH2 termini. The observation of variant cDNA classes suggests that acute phase rat livers produce a heterogeneous mixture of alpha 1-inhibitor III mRNA molecules. Evidence was obtained for the presence of at least four different alpha 1-inhibitor III-related genes in the rat genome. During the first 24 h of an acute phase response the abundance of hepatic alpha 1-inhibitor III mRNA was decreased 3-4-fold. This decrease was of the same order of magnitude as the reported reduction of the corresponding plasma protein concentration, suggesting that in the early phase of the acute inflammatory response the plasma concentration of this protein is mainly controlled through the abundance of its hepatic mRNA.     

Identification of a new P450 expressed in human lung: complete cDNA sequence, cDNA-directed expression, and chromosome mapping

A cDNA coding for a P450 expressed in human lung was isolated from a lambda gt11 library constructed from human lung mRNA using a cDNA probe to rat P450 IVA1. The cDNA-deduced amino acid sequence of this P450, designated IVB1, consisted of 511 amino acids and had a calculated molecular weight of 59,558. The IVB1 amino acid sequence bore 51%, 53%, and 52% similarities to rat IVA1, IVA2, and rabbit P450p-2, respectively. Comparison of the primary amino acid sequence of human IVB1 with rat IVA and rabbit p-2 P450 sequences revealed a region of absolute sequence identity of 17 amino acids between residues 304 and 320. However, the functional significance of this conserved sequence is unknown. Human IVB1 also appears to be related to P450 isozyme 5 that has been extensively characterized in rabbits. The IVB1 cDNA was inserted into a vaccinia virus expression vector and the enzyme expressed in human cell lines. The expressed enzyme had an absorption spectrum with a lambda max at 450 nm when reduced and complexed with carbon monoxide, typical of other cytochrome P450s. Unlike rabbit P450 isozyme 5, however, human IVB1 was unable to activate the promutagen 2-aminofluorene. Human lung microsomal P450s were also unable to metabolize this compound despite the presence of IVB1 mRNA in three out of four human lungs analyzed. In contrast to its expression in lung, IVB1 mRNA was undetectable in livers from 14 individuals, including those from which the lungs were derived. IVB1-related mRNA was also expressed in rat lung and was undetectable in untreated rat liver.(ABSTRACT TRUNCATED AT 250 WORDS)