Organization and nucleotide sequence of the 3' end of the human CAD gene

CAD, A Multienzymatic Protein at the Head of de Novo Pyrimidine Biosynthesis

CAD is a 1.5 MDa particle formed by hexameric association of a 250 kDa protein that carries the enzymatic activities for the first three steps in the de novo biosynthesis of pyrimidine nucleotides: glutamine-dependent Carbamoyl phosphate synthetase, Aspartate transcarbamoylase and Dihydroorotase. This metabolic pathway is essential for cell growth and proliferation and is conserved in all living organisms. However, the fusion of the first three enzymatic activities of the pathway into a single multienzymatic protein only occurs in animals. In prokaryotes, by contrast, these activities are encoded as distinct monofunctional enzymes that function independently or by forming more or less transient complexes. Whereas the structural information about these enzymes in bacteria is abundant, the large size and instability of CAD has only allowed a fragmented characterization of its structure. Here we retrace some of the most significant efforts to decipher the architecture of CAD and to understand its catalytic and regulatory mechanisms.

High rate of CAD gene amplification in human cells deficient in MLH1 or MSH6

MutS and MutL homologs have been implicated in multiple genetic stabilization pathways. The activities participate in the correction of DNA biosynthetic errors, are involved in cellular responses to certain types of DNA damage, and serve to ensure the fidelity of genetic recombination. We show here that the rate of CAD (carbamyl-P synthetase/aspartate transcarbamylase/dihydroorotase) gene amplification is elevated 50- to 100-fold in human cell lines deficient in MLH1 or MSH6, as compared with mismatch repair-proficient control cells. Fluorescence in situ hybridization indicates that these amplification events are the probable consequence of unequal sister chromatid exchanges involving chromosome 2, as well as translocation events involving other chromosomes. These results implicate MutS alpha and MutL alpha in the suppression of gene amplification and suggest that defects in this genetic stabilization function may contribute to the cancer predisposition associated with mismatch repair deficiency.

Imbalanced DNA synthesis induced by cytosine arabinoside and fludarabine in human leukemia cells11Abbreviations: araC, 1-β-d-arabinofuranosylcytosine (cytosine arabinoside); araA, 1-β-d-arabinofuranosyladenine; BrdUrd, 5-bromo-2′-deoxyuridine; FaraA, 1-β-d-arabinofuranosyl-2-fluoroadenine (fludarabine); ic50, concentration that reduces cloning efficiency by 50%; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PALA, N-(phosphonacetyl)-l-aspartate; and SSC, standard saline citrate

Previous studies have demonstrated that cytosine arabinoside (araC) induces an accumulation of Okazaki fragments, while fludarabine (FaraA) inhibits Okazaki fragment synthesis. We extended these observations in the present study to provide insights into various mechanisms by which these anticancer drugs affect DNA replication and induce genomic instability in human CEM leukemia cells. Neither araC nor FaraA induced a detectable amount of re-replicated DNA in S-phase cells, which indicated that drug-induced alterations in Okazaki fragment synthesis were not accompanied by DNA re-replication. Synthesis on both leading and lagging DNA strands within the c-myc locus was measured in cells incubated with equitoxic concentrations of araC or FaraA. In araC-treated cells, nascent DNA from the lagging strand was enriched about 5-fold compared with the leading strand. In contrast, FaraA did not induce any replication imbalance. AraC- and FaraA induced changes in the frequency of N-(phosphonacetyl)-l-aspartate (PALA) resistance and the extent of CAD gene amplification were monitored as markers of drug-induced genomic instability. At concentrations that reduced cloning efficiency by 50% (IC(50)), araC increased the frequency of PALA resistance about 4-fold, while FaraA did not have a significant effect on the frequency of PALA resistance. Pretreatment with araC also increased the extent of CAD gene amplification. We propose that the imbalanced DNA synthesis induced by araC leads to the accumulation of Okazaki fragments on the lagging arms and single-stranded DNA regions on the leading arms of replication forks. The formation of these abnormal replication structures was associated with the generation of genomic instability.

The yeast Ura2 protein that catalyses the first two steps of pyrimidines biosynthesis accumulates not in the nucleus but in the cytoplasm, as shown by immunocytochemistry and Ura2-green fluorescent protein mapping

The Ura2 multidomain protein catalyses the first two steps of pyrimidines biosynthesis in Saccharomyces cerevisiae. It consists of a 240 kDa polypeptide which contains carbamyl phosphate synthetase and aspartate transcarbamylase domains. The Ura2 protein was believed to be nucleoplasmic, since one of the aspartate transcarbamylase reaction products, monophosphate, was reported to be precipitated by lead ions inside nuclei. However, this ultracytochemical approach was recently shown to give artifactual lead polyphosphate precipitates, and the use of cerium instead of lead failed to reveal this nucleoplasmic localization. Ura2 localization has therefore been undertaken by means of three alternative approaches based on the detection of the protein itself: (a) indirect immunofluorescence of yeast protoplasts; (b) immunogold labelling of ultrathin sections of embedded yeast cells (both approaches using affinity purified primary antibodies directed against the 240 kDa Ura2 polypeptide chain, or against a 22 residue peptide specific of the carbamyl phosphate synthetase domain); and (c) direct fluorescence of cells expressing an Ura2-green fluorescent protein hybrid. All three approaches localize the bulk of Ura2 to the cytoplasm, whereas the signals associated with the nucleus, mitochondria or vacuoles are close to or at the background level.

Aspartate-90 and arginine-269 of hamster aspartate transcarbamylase affect the oligomeric state of a chimaeric protein with an Escherichia coli maltose-binding domain

Residues Asp-90 and Arg-269 of Escherichia coli aspartate transcarbamylase seem to interact at the interface of adjacent catalytic subunits. Alanine substitutions at the analogous positions in the hamster aspartate transcarbamylase of a chimaeric protein carrying an E. coli maltose-binding domain lead to changes in both the kinetics of the enzyme and the quaternary structure of the protein. The Vmax for the Asp-90-->Ala and Arg-269-->Ala substitutions is decreased to 1/21 and 1/50 respectively, the [S]0.5 for aspartate is increased 540-fold and 826-fold respectively, and the [S]0.5 for carbamoyl phosphate is increased 60-fold for both. These substitutions decrease the oligomeric size of the protein. Whereas the native chimaeric protein behaves as a pentamer, the Asp-90 variant is a trimer and the Arg-269 variant is a dimer. The altered enzymes also exhibit marked decreases in thermal stability and are inactivated at much lower concentrations of urea than is the unaltered enzyme. Taken together, these results are consistent with the hypothesis that both Asp-90 and Arg-269 have a role in the enzymic function and structural integrity of hamster aspartate transcarbamylase.

Cerium-based ultracytochemical localization of aspartate transcarbamylase activity in the cell membrane complex of Saccharomyces cerevisiae

Aspartate transcarbamylase (ATCase) activity was localized ultracytochemically in the yeast Saccharomyces cerevisiae by precipitation of its reaction product orthophosphate as cerium phosphate. We prefixed yeast cells with ice-cold 1% glutaraldehyde for 30 min which preserved 80% of ATCase activity. Cells were washed and incubated with ATCase substrates (aspartate, carbamyl phosphate) plus cerium chloride, and postfixed by osmium tetroxide. In cells from exponential batch cultures, deposits of cerium phosphate delineated simultaneously or alternatively membranes of the secretory pathway: nuclear envelope, endoplasmic reticulum, Golgi complex and the plasmalemma; mitochondrial membranes and intramitochondrial fibrous component were labelled as well. Deposits of cerium phosphate were never observed in the nucleoplasm. Cells incubated in the absence of cerium or ATCase substrates and mutant S. cerevisiae cells lacking ATCase activity served as controls. Small round electron-dense condensates were found to be randomly distributed within some cells, both in control and experimental runs, in the nucleoplasm, cytoplasm and mitochondrial matrix and represented undefined osmicated endogenous compounds. Our results suggest that the synthesis of pyrimidine precursors occurs in membranes, where compounds such as UDP-glucose and CDP-diglycerides are needed for membrane and/or yeast cell wall synthesis. The possible contribution of ATCase activity found in the nuclear envelope to nucleic acid synthesis remains to be clarified.

Structural modeling and electrostatic properties of aspartate transcarbamylase from Saccharomyces cerevisiae

In Saccharomyces cerevisiae the first two reactions of the pyrimidine pathway are catalyzed by a multifunctional protein which possesses carbamylphosphate synthetase and aspartate transcarbamylase activities. Genetic and proteolysis studies suggested that the ATCase activity is carried out by an independently folded domain. In order to provide structural information for ongoing mutagenesis studies, a model of the three-dimensional structure of this domain was generated on the basis of the known X-ray structure of the related catalytic subunit from E. coli ATCase. First, a model of the catalytic monomer was built and refined by energy minimization. In this structure, the conserved residues between the two proteins were found to constitute the hydrophobic core whereas almost all the mutated residues are located at the surface. Then, a trimeric structure was generated in order to build the active site as it lies at the interface between adjacent chains in the E. coli catalytic trimer. After docking a bisubstrate analog into the active site, the whole structure was energy minimized to regularize the interactions at the contact areas between subunits. The resulting model is very similar to that obtained for the E. coli catalytic trimer by X-ray crystallography, with a remarkable conservation of the structure of the active site and its vicinity. Most of the interdomain and intersubunit interactions that are essential for the stability of the E. coli catalytic trimer are maintained in the yeast enzyme even though there is only 42% identity between the two sequences. Free energy calculations indicate that the trimeric assembly is more stable than the monomeric form. Moreover an insertion of four amino acids is localized in a loop which, in E. coli ATCase, is at the surface of the protein. This insertion exposes hydrophobic residues to the solvent. Interestingly, such an insertion is present in all the eukaryotic ATCase genes sequences so far, suggesting that this region is interacting with another domain of the multifunctional protein.

The propensity for gene amplification: a comparison of protocols, cell lines, and selection agents

We have studied cell lines of rodent and human origin for their propensity to become resistant to antifolates (methotrexate, trimetrexate), phosphonacetyl-L-aspartate (PALA), and colcemid, resistances associated with amplification of the DHFR, CAD, and MDR1 genes, respectively. We have employed two different methods: (1) a shallow step-wise selection protocol, where time to attain specified resistance is the quantitative measure, (2) the frequency of resistant colonies at specified drug concentrations. Although there are advantages and disadvantages to both methods, the two methods gave the same relative ranking of cell lines. Striking differences in the propensity for gene amplification (resistance) were found: human cell lines were less prone to amplify genes than Chinese hamster ovary (CHO) cells. This ranking was similar with all of the agents employed. Additionally, we observed that whereas PALA resistance in CHO cells is associated with amplification of the CAD gene, PALA resistance in the two human cell lines studied (HeLaS3 and VA13) was not associated with amplification and/or overexpression of the CAD gene, and thus this resistance to PALA occurs by an unknown mechanism.

The evolutionary history of the first three enzymes in pyrimidine biosynthesis

Some metabolic pathways are nearly ubiquitous among organisms: the genes encoding the enzymes for such pathways must therefore be ancient and essential. De novo pyrimidine biosynthesis is an example of one such metabolic pathway. In animals a single protein called CAD carries the first three steps of this pathway. The same three enzymes in prokaryotes are associated with separate proteins. The CAD gene appears to have evolved through a process of gene duplication and DNA rearrangement, leading to an in-frame gene fusion encoding a chimeric protein. A driving force for the creation of eukaryotic genes encoding multienzymatic proteins such as CAD may be the advantage of coordinate expression of enzymes catalyzing steps in a biosynthetic pathway. The analogous structure in bacteria is the operon. Differences in the translational mechanisms of eukaryotes and prokaryotes may have dictated the different strategies used by organisms to evolve coordinately regulated genes.

Altered cell cycle arrest and gene amplification potential accompany loss of wild-type p53

Gene amplification occurs at high frequency in transformed cells (10(-3)-10(-5)), but is undetectable in normal diploid fibroblasts (less than 10(-9)). This study examines whether alterations of one or both p53 alleles were sufficient to allow gene amplification to occur. Cells retaining one wild-type p53 allele mimicked the behavior of primary diploid cells: they arrested growth in the presence of drug and failed to demonstrate amplification. Cells losing the second p53 allele failed to arrest when placed in drug and displayed the ability to amplify at a high frequency. Thus, loss of wild-type p53 may lead to amplification, possibly caused by changes in cell cycle progression. Other determinants can by-pass this p53 function, however, since tumor cells with wild-type p53 have the ability to amplify genes.

Wild-type p53 restores cell cycle control and inhibits gene amplification in cells with mutant p53 alleles

Loss of cell cycle control and acquisition of chromosomal rearrangements such as gene amplification often occur during tumor progression, suggesting that they may be correlated. We show here that the wild-type p53 allele is lost when fibroblasts from patients with the Li-Fraumeni syndrome (LFS) are passaged in vitro. Normal and LFS cells containing wild-type p53 arrested in G1 when challenged with the uridine biosynthesis inhibitor PALA and did not undergo PALA-selected gene amplification. The converse occurred in cells lacking wild-type p53 expression. Expression of wild-type p53 in transformants of immortal and tumor cells containing mutant p53 alleles restored G1 control and reduced the frequency of gene amplification to undetectable levels. These studies reveal that p53 contributes to a metabolically regulated G1 check-point, and they provide a model for understanding how abnormal cell cycle progression leads to the genetic rearrangements involved in tumor progression.

A single base change at a splice acceptor site leads to a truncated CAD protein in Urd-A mutant Chinese hamster ovary cells

We have previously reported the isolation and characterization of mutant Chinese hamster ovary (CHO-K1) cells of the Urd-A complementation group, which require uridine for growth, are deficient in the activities of the first three enzymes of de novo UMP biosynthesis, and produce markedly reduced amounts of a truncated form of the multifunctional protein CAD, which contains these three enzyme activities. We report here that a single base change of G to A at a highly conserved RNA splice acceptor site is responsible for the phenotype of this mutant. In addition to a small amount of apparently normal CAD mRNA, this mutation causes production of two alternative forms of CAD mRNA in the mutant, one that includes the intron just prior to the mutation and one that excludes the exon just after the mutation. The affected splice site is located at the intron-exon boundary just preceding the exon that encodes the beginning of the aspartate transcarbamylase (ATCase) domain of the CAD protein. Both intron inclusion and exon exclusion during RNA processing introduce a translation stop codon upstream of the region encoding this domain, resulting in the production of the truncated CAD protein seen in the Urd-A mutant. This mutation also results in markedly decreased levels of CAD mRNA and protein in the mutant.

An unusual Alu repeat sequence within the CAD gene

There are several hundred thousand members of the Alu repeat family in the human genome. Those Alu elements sequenced to date appear to fit into subfamilies. A novel Alu has been found in an intron of the human CAD gene: it appears to be due to rearrangement between Alu repeats belonging to two different subfamilies. Further sequence data from this intron suggest that the Alu element may have rearranged prior to its entry into the CAD gene. Such findings indicate that, in addition to single nucleotide substitutions and deletions, DNA rearrangements may be a factor in generating the diversity of Alu repeats found in primate genomes.