Nuclear and mitochondrial forms of human uracil-DNA glycosylase are encoded by the same gene.

Recent cloning of a cDNA (UNG15) encoding human uracil-DNA glycosylase (UDG), indicated that the gene product of M(r) = 33,800 contains an N-terminal sequence of 77 amino acids not present in the presumed mature form of M(r) = 25,800. This led to the hypothesis that the N-terminal sequence might be involved in intracellular targeting. To examine this hypothesis, we analysed UDG from nuclei, mitochondria and cytosol by western blotting and high resolution gel filtration. An antibody that recognises a sequence in the mature form of the UNG protein detected all three forms, indicating that they are products of the same gene. The nuclear and mitochondrial form had an apparent M(r) = 27,500 and the cytosolic form an apparent M(r) = 38,000 by western blotting. Gel filtration gave essentially similar estimates. An antibody with specificity towards the presequence recognised the cytosolic form of M(r) = 38,000 only, indicating that the difference in size is due to the presequence. Immunofluorescence studies of HeLa cells clearly demonstrated that the major part of the UDG activity was localised in the nuclei. Transfection experiments with plasmids carrying full-length UNG15 cDNA or a truncated form of UNG15 encoding the presumed mature UNG protein demonstrated that the UNG presequence mediated sorting to the mitochondria, whereas UNG lacking the presequence was translocated to the nuclei. We conclude that the same gene encodes nuclear and mitochondrial uracil-DNA glycosylase and that the signals for mitochondrial translocation resides in the presequence, whereas signals for nuclear import are within the mature protein.     

Mitochondrial base excision repair of uracil and AP sites takes place by single-nucleotide insertion and long-patch DNA synthesis

Base excision repair (BER) corrects a variety of small base lesions in DNA. The UNG gene encodes both the nuclear (UNG2) and the mitochondrial (UNG1) forms of the human uracil-DNA glycosylase (UDG). We prepared mitochondrial extracts free of nuclear BER proteins from human cell lines. Using these extracts we show that UNG is the only detectable UDG in mitochondria, and mitochondrial BER (mtBER) of uracil and AP sites occur by both single-nucleotide insertion and long-patch repair DNA synthesis. Importantly, extracts of mitochondria carry out repair of modified AP sites which in nuclei occurs through long-patch BER. Such lesions may be rather prevalent in mitochondrial DNA because of its proximity to the electron transport chain, the primary site of production of reactive oxygen species. Furthermore, mitochondrial extracts remove 5' protruding flaps from DNA which can be formed during long-patch BER, by a "flap endonuclease like" activity, although flap endonuclease (FEN1) is not present in mitochondria. In conclusion, combined short- and long-patch BER activities enable mitochondria to repair a broader range of lesions in mtDNA than previously known.     

Identification,cloning, and expression of uracil-DNA glycosylase from Atlantic cod (<Emphasis Type="Italic">Gadus morhua</Emphasis>): characterization and homology modeling of the cold-active catalytic domain

Two distinct forms of the highly conserved uracil-DNA glycosylase (UNG) have been isolated from Atlantic cod (Gadus morhua) liver cDNA by rapid amplification of cDNA ends (RACE). From the cDNA sequences, both forms were deduced to encode an open reading frame of 301 amino acids, with an identical 267-amino-acid C-terminal region and different N-terminal regions of 34 amino acids. By comparison with the human UNG sequences, the two forms were identified as possible mitochondrial (cUNG1) and nuclear (cUNG2) forms. Several constructs of recombinant cUNG (rcUNG) were expressed in Escherichia coli in order to optimize the yield. The recombinant enzyme was purified to apparent homogeneity as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Activity and stability experiments showed that rcUNG was similar to cUNG previously purified from Atlantic cod liver, and was more pH- and temperature labile than a recombinant human UNG (rhUNG). Under optimal assay conditions for both rcUNG and rhUNG, the turnover number (k(cat)) was three times higher for rcUNG compared with rhUNG, with an identical K(M), resulting in a threefold higher catalytic efficiency (k(cat)/K(M)) for rcUNG. These activity and stability experiments reveal cold-adapted features in rcUNG. Homology models of the catalytic domains of Atlantic cod (cUNG) and mouse uracil-DNA glycosylase (mUNG) were built using the human UNG (hUNG) crystal structure as a template. The unique amino acid substitutions observed in cod UNG were mainly located in the N- and C-terminal parts of the sequence. The analysis indicated a more stable N-terminal, a more flexible C-terminal, and a less stabilized core in cUNG as compared with the mammalian UNGs. Substitution of several amino acids in or near the DNA-binding site in cUNG could give rise to a more positively charged surface and a higher electrostatic potential near the active site compared with the mammalian UNGs. The higher potential may increase the electrostatic interactions between the enzyme and DNA, and may explain the increased substrate affinity and, in combination with the higher flexibility, the higher catalytic efficiency observed for rcUNG.     

The rate of base excision repair of uracil is controlled by the initiating glycosylase

Uracil in DNA is repaired by base excision repair (BER) initiated by a DNA glycosylase, followed by strand incision, trimming of ends, gap filling and ligation. Uracil in DNA comes in two distinct forms; U:A pairs, typically resulting from replication errors, and mutagenic U:G mismatches, arising from cytosine deamination. To identify proteins critical to the rate of repair of these lesions, we quantified overall repair of U:A pairs, U:G mismatches and repair intermediates (abasic sites and nicked abasic sites) in vitro. For this purpose we used circular DNA substrates and nuclear extracts of eight human cell lines with wide variation in the content of BER proteins. We identified the initiating uracil-DNA glycosylase UNG2 as the major overall rate-limiting factor. UNG2 is apparently the sole glycosylase initiating BER of U:A pairs and generally initiated repair of almost 90% of the U:G mismatches. Surprisingly, TDG contributed at least as much as single-strand selective monofunctional uracil-DNA glycosylase 1 (SMUG1) to BER of U:G mismatches. Furthermore, in a cell line that expressed unusually high amounts of TDG, this glycosylase contributed to initiation of as much as approximately 30% of U:G repair. Repair of U:G mismatches was generally faster than that of U:A pairs, which agrees with the known substrate preference of UNG-type glycosylases. Unexpectedly, repair of abasic sites opposite G was also generally faster than when opposite A, and this could not be explained by the properties of the purified APE1 protein. It may rather reflect differences in substrate recognition or repair by different complex(es). Lig III is apparently a minor rate-regulator for U:G repair. APE1, Pol beta, Pol delta, PCNA, XRCC1 and Lig I did not seem to be rate-limiting for overall repair of any of the substrates. These results identify damaged base removal as the major rate-limiting step in BER of uracil in human cells.     

A functionally divergent hydrogenosomal peptidase with protomitochondrial ancestry

Matrix proteins of mitochondria, hydrogenosomes and mitosomes are typically targeted and translocated into their respective organelles using N-terminal presequences that are subsequently cleaved by a peptidase. Here we characterize a approximately 47 kDa metallopeptidase, from the hydrogenosome-bearing, unicellular eukaryote Trichomonas vaginalis, that contains the active site motif (HXXEHX(76)E) characteristic of the beta subunit of the mitochondrial processing peptidase (MPP) and localizes to hydrogenosomes. The purified recombinant protein, named hydrogenosomal processing peptidase (HPP), is capable of cleaving a hydrogenosomal presequence in vitro, in contrast to MPP which requires both an alpha and beta subunit for activity. T. vaginalis HPP forms an approximately 100 kDa homodimer in vitro and also exists in an approximately 100 kDa complex in vivo. Our phylogenetic analyses support a common origin for HPP and betaMPP and demonstrate that gene duplication gave rise to alphaMPP and betaMPP before the divergence of T. vaginalis and mitochondria-bearing lineages. These data, together with published analyses of MPPs and putative mitosomal processing peptidases, lead us to propose that the length of targeting presequences and the subunit composition of organellar processing peptidases evolved in concert. Specifically, longer mitochondrial presequences may have evolved to require an alpha/beta heterodimer for accurate cleavage, while shorter hydrogenosomal and mitosomal presequences did not.     

Affinity Purification and Comparative Analysis of Two Distinct Human Uracil-DNA Glycosylases

Evidence is presented on two forms of uracil-DNA glycosylase (UDG1 and UDG2) that exist in human cells. We have developed an affinity technique to isolate uracil-DNA glycosylases from HeLa cells. This technique relies on the use of a uracil-DNA glycosylase inhibitor (Ugi) produced by theBacillus subtilisbacteriophage, PBS2. Affinity-purified preparations of uracil-DNA glycosylase, derived from total HeLa cell extracts, reveal a group of bands in the 36,000 molecular weight range and a single 30,000 molecular weight band when analyzed by SDS–PAGE and silver staining. In contrast, only the 30,000 molecular weight band is seen in HeLa mitochondrial preparations. Separation of HeLa cell nuclei from the postnuclear supernatant reveals that uracil-DNA glycosylase activity is evenly distributed between the nuclear compartment and the postnuclear components of the cell. Immunostaining of a nuclear extract with antisera to UDG1 indicates that the nuclear associated uracil-DNA glycosylase activity is not associated with the highly conserved uracil-DNA glycosylase, UDG1. With the use of Ugi-Sepharose affinity chromatography, we show that a second and distinct uracil-DNA glycosylase is associated with the nuclear compartment. Immunoblot analysis, utilizing antisera generated against UDG1, reveals that the 30,000 molecular weight protein and a protein in the 36,000 range share common epitopes. Cycloheximide treatment of HeLa cells indicates that upon inhibition of protein synthesis, the higher molecular weight species disappears and is apparently posttranslationally processed into a lower molecular weight form. This is substantiated by mitochondrial import studies which reveal thatin vitroexpressed UDG1 becomes resistant to trypsin treatment within 15 min of incubation with mitochondria. Within this time frame, a lower molecular weight form of uracil-DNA glycosylase appears and is associated with the mitochondria. Antibodies generated against peptides from specific regions of the cyclin-like uracil-DNA glycosylase (UDG2), demonstrate that this nuclear glycosylase is a phosphoprotein with a molecular weight in the range of 36,000. SDS–PAGE analysis of Ugi affinity-purified and immunoprecipitated UDG2 reveals two closely migrating phosphate-containing species, indicating that UDG2 either contains multiple phosphorylation sites (resulting in heterogeneous migration) or that two distinct forms of UDG2 exist in the cell. Cell staining of various cultured human cell lines corroborates the finding that UDG1 is largely excluded from the nucleus and that UDG2 resides mainly in the nucleus. Our results indicate that UDG1 is targeted to the mitochondria and undergoes proteolytic processing typical of resident mitochondrial proteins that are encoded by nuclear DNA. These results also indicate that the cyclin-like uracil-DNA glycosylase (UDG2) may be a likely candidate for the nuclear located base-excision repair enzyme.     

Uracil-DNA glycosylase (UNG)-deficient mice reveal a primary role of the enzyme during DNA replication

Gene-targeted knockout mice have been generated lacking the major uracil-DNA glycosylase, UNG. In contrast to ung- mutants of bacteria and yeast, such mice do not exhibit a greatly increased spontaneous mutation frequency. However, there is only slow removal of uracil from misincorporated dUMP in isolated ung-/- nuclei and an elevated steady-state level of uracil in DNA in dividing ung-/- cells. A backup uracil-excising activity in tissue extracts from ung null mice, with properties indistinguishable from the mammalian SMUG1 DNA glycosylase, may account for the repair of premutagenic U:G mispairs resulting from cytosine deamination in vivo. The nuclear UNG protein has apparently evolved a specialized role in mammalian cells counteracting U:A base pairs formed by use of dUTP during DNA synthesis.     

Excision of deaminated cytosine from the vertebrate genome: role of the SMUG1 uracil-DNA glycosylase

Gene-targeted mice deficient in the evolutionarily conserved uracil-DNA glycosylase encoded by the UNG gene surprisingly lack the mutator phenotype characteristic of bacterial and yeast ung(-) mutants. A complementary uracil-DNA glycosylase activity detected in ung(-/-) murine cells and tissues may be responsible for the repair of deaminated cytosine residues in vivo. Here, specific neutralizing antibodies were used to identify the SMUG1 enzyme as the major uracil-DNA glycosylase in UNG-deficient mice. SMUG1 is present at similar levels in cell nuclei of non-proliferating and proliferating tissues, indicating a replication- independent role in DNA repair. The SMUG1 enzyme is found in vertebrates and insects, whereas it is absent in nematodes, plants and fungi. We propose a model in which SMUG1 has evolved in higher eukaryotes as an anti-mutator distinct from the UNG enzyme, the latter being largely localized to replication foci in mammalian cells to counteract de novo dUMP incorporation into DNA.     

Rat liver mitochondrial intermediate peptidase (MIP): purification and initial characterization.

A number of nuclearly encoded mitochondrial protein precursors that are transported into the matrix and inner membrane are cleaved in two sequential steps by two distinct matrix peptidases, mitochondrial processing peptidase (MPP) and mitochondrial intermediate peptidase (MIP). We have isolated and purified MIP from rat liver mitochondrial matrix. The enzyme, purified 2250-fold, is a monomer of 75 kDa and cleaves all tested mitochondrial intermediate proteins to their mature forms. About 20% of the final MIP preparation consists of equimolar amounts of two peptides of 47 kDa and 28 kDa, which are apparently the products of a single cleavage of the 75 kDa protein. These peptides are not separable from the 75 kDa protein, nor from each other, under any conditions used in the purification. The peptidase has a broad pH optimum between pH 6.6 and 8.9 and is inactivated by N-ethylmaleimide (NEM) and other sulfhydryl group reagents. The processing activity is divalent cation-dependent; it is stimulated by manganese, magnesium or calcium ions and reversibly inhibited by EDTA. Zinc, cobalt and iron strongly inhibit MIP activity. This pattern of cation dependence and inhibition is not clearly consistent with that of any known family of proteases.     

Purification of the human O6-methylguanine-DNA methyltransferase and uracil-DNA glycosylase, the latter to apparent homogeneity

Uracil-DNA glycosylase, the enzyme that catalyzes the release of free uracil from single-stranded and double-stranded DNA, has been purified 26,600-fold from HeLa S3 cell extracts. The enzyme preparation was essentially homogeneous as judged by sodium dodecyl sulfate/polyacrylamide gel electrophoresis. The native enzyme is a small monomeric protein of molecular mass 29 kDa. A minor uracil-DNA glycosylase preparation was also obtained in the final chromatographic step. This preparation is homogeneous with a molecular mass of 29 kDa and may represent the mitochondrial enzyme. This report also presents a 700-fold purification of HeLa S3 cell O6-methylguanine-DNA methyltransferase. The glycosylase and methyltransferase showed very similar chromatographic properties. The report indicates that the lability of the methyltransferase upon purification may be a consequence of the total separation of the two DNA repair enzymes or of the possibility that some other stabilizing factor is involved.     

Mitochondrial protein turnover: role of the precursor intermediate peptidase Oct1 in protein stabilization

Most mitochondrial proteins are encoded in the nucleus as precursor proteins and carry N-terminal presequences for import into the organelle. The vast majority of presequences are proteolytically removed by the mitochondrial processing peptidase (MPP) localized in the matrix. A subset of precursors with a characteristic amino acid motif is additionally processed by the mitochondrial intermediate peptidase (MIP) octapeptidyl aminopeptidase 1 (Oct1), which removes an octapeptide from the N-terminus of the precursor intermediate. However, the function of this second cleavage step is elusive. In this paper, we report the identification of a novel Oct1 substrate protein with an unusual cleavage motif. Inspection of the Oct1 substrates revealed that the N-termini of the intermediates typically carry a destabilizing amino acid residue according to the N-end rule of protein degradation, whereas mature proteins carry stabilizing N-terminal residues. We compared the stability of intermediate and mature forms of Oct1 substrate proteins in organello and in vivo and found that Oct1 cleavage increases the half-life of its substrate proteins, most likely by removing destabilizing amino acids at the intermediate's N-terminus. Thus Oct1 converts unstable precursor intermediates generated by MPP into stable mature proteins.     

Mutations at Arginine 276 transform human uracil-DNA glycosylase into a single-stranded DNA-specific uracil-DNA glycosylase

To investigate the role of Arginine 276 in the conserved leucine-loop of human uracil-DNA glycosylase (UNG), the effects of six R276 amino acid substitutions (C, E, H, L, W, and Y) on nucleotide flipping and enzyme conformational change were determined using transient and steady state, fluorescence-based, kinetic analysis. Relative to UNG, the mutant proteins exhibited a 2.6- to 7.7-fold reduction in affinity for a doubled-stranded oligonucleotide containing a pseudouracil residue opposite 2-aminopurine, as judged by steady-state DNA binding-base flipping assays. An anisotropy binding assay was utilized to determine the K(d) of UNG and the R276 mutants for carboxyfluorescein-labeled uracil-containing single- and double-stranded oligonucleotides; the binding affinities varied 11-fold for single-stranded uracil-DNA, and 43-fold for double-stranded uracil-DNA. Productive uracil-DNA binding was monitored by rapid quenching of UNG intrinsic protein fluorescence. Relative to UNG, the rate of intrinsic fluorescence quenching of five mutant proteins for binding double-stranded uracil-DNA was reduced approximately 50%; the R276E mutant exhibited 1% of the rate of fluorescence quenching of UNG. When reacted with single-stranded uracil-DNA, the rate of UNG fluorescence quenching increased. Moreover, the rate of fluorescence quenching for all the mutant proteins, except R276E, was slightly faster than UNG. The k(cat) of the R276 mutants was comparable to UNG on single-stranded DNA and differentially affected by NaCl; however, k(cat) on double-stranded DNA substrate was reduced 4-12-fold and decreased sharply at NaCl concentrations as low as 20 mM. Taken together, these results indicate that the effects of mutations at Arg276 were largely limited to enzyme interactions with double-stranded uracil-containing DNA, and suggested that mutations at Arg276 effectively transformed UNG into a single-stranded DNA-specific uracil-DNA glycosylase.     

Processing of the dual targeted precursor protein of glutathione reductase in mitochondria and chloroplasts

Pea glutathione reductase (GR) is dually targeted to mitochondria and chloroplasts by means of an N-terminal signal peptide of 60 amino acid residues. After import, the signal peptide is cleaved off by the mitochondrial processing peptidase (MPP) in mitochondria and by the stromal processing peptidase (SPP) in chloroplasts. Here, we have investigated determinants for processing of the dual targeting signal peptide of GR by MPP and SPP to examine if there is separate or universal information recognised by both processing peptidases. Removal of 30 N-terminal amino acid residues of the signal peptide (GRDelta1-30) greatly stimulated processing activity by both MPP and SPP, whereas constructs with a deletion of an additional ten amino acid residues (GRDelta1-40) and deletion of 22 amino acid residues in the middle of the GR signal sequence (GRDelta30-52) could be cleaved by SPP but not by MPP. Numerous single mutations of amino acid residues in proximity of the cleavage site did not affect processing by SPP, whereas mutations within two amino acid residues on either side of the processing site had inhibitory effect on processing by MPP with a nearly complete inhibition for mutations at position -1. Mutation of positively charged residues in the C-terminal half of the GR targeting peptide inhibited processing by MPP but not by SPP. An inhibitory effect on SPP was detected only when double and triple mutations were introduced upstream of the cleavage site. These results indicate that: (i) recognition of processing site on a dual targeted GR precursor differs between MPP and SPP; (ii) the GR targeting signal has similar determinants for processing by MPP as signals targeting only to mitochondria; and (iii) processing by SPP shows a low level of sensitivity to single mutations on targeting peptide and likely involves recognition of the physiochemical properties of the sequence in the vicinity of cleavage rather than a requirement for specific amino acid residues.     

The UNG2 Arg88Cys variant abrogates RPA-mediated recruitment of UNG2 to single-stranded DNA

In human cell nuclei, UNG2 is the major uracil-DNA glycosylase initiating DNA base excision repair of uracil. In activated B cells it has an additional role in facilitating mutagenic processing of AID-induced uracil at Ig loci and UNG-deficient patients develop hyper-IgM syndrome characterized by impaired class-switch recombination and disturbed somatic hypermutation. How UNG2 is recruited to either error-free or mutagenic uracil processing remains obscure, but likely involves regulated interactions with other proteins. The UNG2 N-terminal domain contains binding motifs for both proliferating cell nuclear antigen (PCNA) and replication protein A (RPA), but the relative contribution of these interactions to genomic uracil processing is not understood. Interestingly, a heterozygous germline single-nucleotide variant leading to Arg88Cys (R88C) substitution in the RPA-interaction motif of UNG2 has been observed in humans, but with unknown functional relevance. Here we demonstrate that UNG2-R88C protein is expressed from the variant allele in a lymphoblastoid cell line derived from a heterozygous germ line carrier. Enzyme activity as well as localization in replication foci of UNG2-R88C was similar to that of WT. However, binding to RPA was essentially abolished by the R88C substitution, whereas binding to PCNA was unaffected. Moreover, we show that disruption of the PCNA-binding motif impaired recruitment of UNG2 to S-phase replication foci, demonstrating that PCNA is a major factor for recruitment of UNG2 to unperturbed replication forks. Conversely, in cells treated with hydroxyurea, RPA mediated recruitment of UNG2 to stalled replication forks independently of functional PCNA binding. Modulation of PCNA- versus RPA-binding may thus constitute a functional switch for UNG2 in cells subsequent to genotoxic stress and potentially also during the processing of uracil at the immunoglobulin locus in antigen-stimulated B cells.     

Human uracil-DNA glycosylase complements E. coli ung mutants.

We have previously isolated a cDNA encoding a human uracil-DNA glycosylase which is closely related to the bacterial and yeast enzymes. In vitro expression of this cDNA produced a protein with an apparent molecular weight of 34 K in agreement with the size predicted from the sequence data. The in vitro expressed protein exhibited uracil-DNA glycosylase activity. The close resemblance between the human and the bacterial enzyme raised the possibility that the human enzyme may be able to complement E. coli ung mutants. In order to test this hypothesis, the human uracil-DNA glycosylase cDNA was established in a bacterial expression vector. Expression of the human enzyme as a LacZ alpha-humUNG fusion protein was then studied in E. coli ung mutants. E. coli cells lacking uracil-DNA glycosylase activity exhibit a weak mutator phenotype and they are permissive for growth of phages with uracil-containing DNA. Here we show that the expression of human uracil-DNA glycosylase in E. coli can restore the wild type phenotype of ung mutants. These results demonstrate that the evolutionary conservation of the uracil-DNA glycosylase structure is also reflected in the conservation of the mechanism for removal of uracil from DNA.     

Repair of U/G and U/A in DNA by UNG2-associated repair complexes takes place predominantly by short-patch repair both in proliferating and growth-arrested cells

Nuclear uracil-DNA glycosylase UNG2 has an established role in repair of U/A pairs resulting from misincorporation of dUMP during replication. In antigen-stimulated B-lymphocytes UNG2 removes uracil from U/G mispairs as part of somatic hypermutation and class switch recombination processes. Using antibodies specific for the N-terminal non-catalytic domain of UNG2, we isolated UNG2-associated repair complexes (UNG2-ARC) that carry out short-patch and long-patch base excision repair (BER). These complexes contain proteins required for both types of BER, including UNG2, APE1, POLbeta, POLdelta, XRCC1, PCNA and DNA ligase, the latter detected as activity. Short-patch repair was the predominant mechanism both in extracts and UNG2-ARC from proliferating and less BER-proficient growth-arrested cells. Repair of U/G mispairs and U/A pairs was completely inhibited by neutralizing UNG-antibodies, but whereas added recombinant SMUG1 could partially restore repair of U/G mispairs, it was unable to restore repair of U/A pairs in UNG2-ARC. Neutralizing antibodies to APE1 and POLbeta, and depletion of XRCC1 strongly reduced short-patch BER, and a fraction of long-patch repair was POLbeta dependent. In conclusion, UNG2 is present in preassembled complexes proficient in BER. Furthermore, UNG2 is the major enzyme initiating BER of deaminated cytosine (U/G), and possibly the sole enzyme initiating BER of misincorporated uracil (U/A).