Definitive Identification of Mammalian 5-Hydroxymethyluracil DNA N -Glycosylase Activity as SMUG1*

Purification from calf thymus of a DNA N -glycosylase activity (HMUDG) that released 5-hydroxymethyluracil (5hmUra) from the DNA of Bacillus subtilis phage SPO1 was undertaken. Analysis of the most purified fraction by SDS-polyacrylamide gel electrophoresis revealed a multiplicity of protein species making it impossible to identify HMUDG by inspection. Therefore, we renatured the enzyme after SDS-polyacrylamide gel electrophoresis and assayed slices of the gel for DNA N -glycosylase activity directed against 5hmUra. Maximum enzymatic activity was identified between molecular mass markers 30 and 34 kDa. Protein was extracted from gel slices and subjected to tryptic digestion and analysis by mass spectrometry. Analysis revealed the presence of 11 peptides that were homologous or identical to the sequence of the recently characterized human single-stranded monofunctional uracil DNA N -glycosylase (hSMUG1). The cDNA of hSMUG1 was isolated and expressed as a recombinant glutathione S -transferase fusion protein that was shown to release 5hmUra with 20 (cid:1) the specific activity of the most purified bovine fraction. We conclude that hSMUG1 and

5-Hydroxymethyluracil (5hmUra) 1 is formed in DNA exposed to ionizing radiation both in solution and in cells. Its formation has been posited to be caused by the indirect action of ionizing radiation under aerobic conditions via the hydroxyl radical-mediated abstraction of a hydrogen radical from the 5-methyl-moiety of thymine followed by oxidation via molecular oxygen (1). 5hmUra has also been identified in human urine and in the DNA of white cells, and the amount has been found to vary under differing conditions of oxidative stress (2)(3)(4).
Human autoantibodies directed against 5-hydroxymethyl-2Јdeoxyuridine in DNA have been identified, and the titer has been correlated with relative risk for certain types of cancer (5).
Despite identification as a radiation product nearly 20 years ago, the biological consequences of the formation of 5hmUra from thymine in DNA remain uncertain. 5hmUra is a normal constituent of SPO1 Bacillus subtilis phage DNA (6), functioning as a coding partner for adenine in place of thymine, and is present in large amounts in the DNA of dinoflagellates (7). Nonetheless, 5hmUra is a substrate for a DNA N-glycosylase activity, 5hmUra DNA N-glycosylase (HMUDG), which releases 5hmUra from the DNA backbone (8,9). Compared with most other DNA N-glycosylases, HMUDG activity is unusual in that no comparable activity has been detected in bacteria (10) or yeast (9).
We undertook a study of the phylogenetic distribution of HMUDG activity (11), and our results indicate that the phylogenetic distribution of HMUDG was linked to the use by the organism of 5-methylcytosine (5meCyt) as a mediator of gene expression. We suggested that 5hmUra could be formed in DNA via two pathways. One pathway was the oxidation of the 5-methyl group of thymine yielding 5hmUra directly. Alternatively, 5hmUra may arise via oxidation of the 5-methyl group of 5meCyt yielding 5hmCyt followed by deamination of 5hmCyt to 5hmUra (12). The propensity of 5hmCyt residues to deaminate to 5hmUra has long ago been observed in the DNA of the Escherichia coli T-even phages. These phages utilize glucosylated 5hmCyt as a normal constituent of DNA and exhibit a very high rate of C to T transitions upon heating (13). Thus, in cellular DNA the oxidation and subsequent deamination of 5meCyt, if left unrepaired, may result in a C to T transition in CpG islands.
We previously reported the partial purification of HMUDG from calf thymus and demonstrated that HMUDG activity was distinct from the major uracil DNA N-glycosylase (UDG) activity of the cell (12). However, we have been unable to achieve a degree of purification sufficient to identify the HMUDG protein species with certainty from examination of an SDS-PAGE analysis of our most purified calf thymus fraction. In this report we describe the methods by which we were able to unequivocally demonstrate that HMUDG is the recently characterized singlestranded monofunctional uracil DNA N-glycosylase (SMUG1) (14).

EXPERIMENTAL PROCEDURES
Purification of HMUDG from Calf Thymus-Fresh calf thymus obtained from local abattoirs, generally weighing between 1.2 and 2 kg, was trimmed of connective tissue. 600-g portions were homogenized in a Waring Blendor together with 3 volumes of Buffer A (25 mM HEPES pH 7.4, 7.5 mM EDTA, 1 mM dithiothreitol, and 100 l of protease inhibitor mixture (Sigma) per 1.0 liter of buffer). The total volume of * This work was supported by National Institutes of Health Grants CA 16669 and CA 49869 (to G. W. T.), CA 16087 (to Kaplan Comprehensive Cancer Center) and 5T32 CA-09161 (to D. R. M.), and a Whitehead Presidential Fellowship (to T. A. N.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
homogenate from a typical preparation was between 5 and 7 liters.
To this homogenate 5 M NaCl was added to a final concentration of 320 mM NaCl. The resulting gelatinous precipitate of nucleoprotein was removed by spooling onto a 10-ml glass pipette. The remaining liquid was filtered through cheesecloth followed by clearing via centrifugation in the GS-3 rotor of a Sorvall RC-5B centrifuge at 9000 rpm for 20 min to yield Fraction I (15).
Fraction I was precipitated with solid (NH 4 ) 2 SO 4 to a final concentration of 70%, the pH continually being adjusted with NH 4 OH to maintain pH 7.5. After centrifugation, the pellets were washed with 100% (NH 4 ) 2 SO 4 and stored at 4°C in aliquots representing the pelleted material from 500 ml of the original 70% (NH 4 ) 2 SO 4 suspension, yielding Fraction II.
Fraction II was dialyzed overnight against Buffer A and applied to a 100-ml bed cation exchange resin in an XK 26/20 column (SP Fastflow, Amersham Pharmacia Biotech). Aliquots of Fraction II applied to the column averaged 3-4 g of protein/sample, which was half the maximal binding capacity of the resin. The column was washed with Buffer A until absorbance of the eluent at 280 nm was at a minimum. Protein was eluted with a 1-liter gradient of Buffer A increasing in NaCl concentration from 0 to 1 M at a flow rate of 5 ml/min. Fractions were collected as 10-ml aliquots and were assayed for HMUDG activity as described below. Fractions containing HMUDG activity were pooled as Fraction III. Fraction III was adjusted to 70% (NH 4 ) 2 SO 4 by the addition of solid (NH 4 ) 2 SO 4 and then pelleted by centrifugation to make Fraction IV, which was stored in aliquots at 4°C.
Aliquots of Fraction IV, dissolved in Buffer A in 400 mM NaCl, were applied to a Hiload chromatography Superdex-75 preparative column that had been calibrated with the gel filtration low molecular weight calibration kit (Amersham Pharmacia Biotech). The column was eluted at a flow rate of 2.5 ml/min using 400 mM NaCl Buffer A, and fractions were collected as 5-ml aliquots. Fifty microliters of each aliquot were assayed for HMUDG activity, and aliquots containing HMUDG activity were pooled as Fraction V.
Fraction V was pooled and concentrated using a Mono-S column (Amersham Pharmacia Biotech, HR10/10) equilibrated with Buffer A in 20 mM NaCl. The protein was eluted with a 20 mM to 1 M NaCl 100-ml gradient of Buffer A and collected as 1-ml aliquots. Aliquots containing HMUDG activity were pooled as Fraction VI.
Fraction VI, which was ϳ200 mM with respect to NaCl as determined by measurement of conductance, was diluted 10-fold to 20 mM NaCl with Buffer A and then applied to an 80-ml heparin-Sepharose CL-6B column (Amersham Pharmacia Biotech). The protein was eluted with a 20 mM to 1 M NaCl 800-ml gradient of Buffer A at a flow rate of 2 ml/min and collected as 8-ml aliquots. Aliquots containing HMUDG activity were pooled as Fraction VII.
Fraction VII was pooled and dialyzed in 20 mM NaCl Buffer A and applied to a 5-ml single-stranded DNA cellulose column (Sigma). The column was pre-equilibrated with dialysis buffer. Protein was eluted with a 20 mM to 1 M NaCl 50-ml gradient of Buffer A at a flow rate of 0.5 ml/min, and 1-ml aliquots were collected. Aliquots containing HMUDG activity were pooled as Fraction VIII.
Assay of HMUDG Activity-[ 3 H]SPO1 phage DNA, prepared as described previously, was used as substrate (12). 3 mCi of [5,6-3 H]uridine (PerkinElmer Life Sciences) were added to a phage-infected 100-ml log phase bacterial culture. This yielded between 30 and 50 g of phage DNA of a specific radioactivity of between 10 5 and 2 ϫ 10 5 cpm/g as measured by A 260 (1 OD unit is 50 g of double-stranded DNA). Enzymatic digestion of SPO1 DNA to 2Ј-deoxyribonucleosides followed by HPLC analysis attached to an on-line liquid scintillation counter revealed 65-70% of the radioactivity to be in Cyt moieties and 30 -35% to be in 5hmUra moieties. Non-radioactive Cyt and 5hmUra (Sigma) were used as UV-absorbing migration standards (12,16).
Assays of HMUDG activity were performed in 200-l volumes in Buffer A (12). The reactions were terminated after 30 or 60 min, and the macromolecules were precipitated with 100 l of BSA (50 mg/ml) and 5 ml of ice-cold acetone. After standing for 15 min at 4°C the tubes were centrifuged in a Beckmann 6KR centrifuge at 4000 rpm for 15 min and the supernatant acetone decanted. The acetone was evaporated under the hood, and the dried residue was redissolved in 500 l of H 2 O. The radioactive content of the aqueous solution containing the redissolved residue was measured in a liquid scintillation counter. The specificity of the HMUDG reaction was confirmed by HPLC analysis coupled with an on-line liquid scintillation counter, of an aliquot of the aqueous supernatant using non-radioactive 5hmUra as a UV-absorbing migration standard (12).
Purification of HMUDG by SDS-PAGE Followed by Renaturation of the Enzyme-Electrophoresis was performed according to the procedures of Hilbert et al. (15) with minor modifications. The protein sample to be analyzed was boiled for 10 min and then loaded and run on a 13% gel (5% stacking gel) with low range molecular size markers (Bio-Rad) at 80 mA. Samples were applied to duplicate lanes to allow one lane to be used for mass spectrometric analysis. After electrophoresis the gel was divided, and the portion to be used for molecular weight alignment was stained with Coomassie Blue. The unstained portion was kept in buffer. The marker gel was then destained and soaked in doubledistilled H 2 O to allow it to expand to its original size. The stained and unstained gels were aligned on a sheet of parafilm. The unstained gel was cut into 0.4-cm slices. One-third of each slice was reserved for mass spectrometry, and the remainder of each slice was placed into 1.5-ml Eppendorf tubes with 400 l of elution buffer (25 mM HEPES pH 7.9, 0.5 mM EDTA, 50 mM KCl, 1 mM dithiothreitol, and 0.1 mg/ml acetylated BSA).
The following renaturation procedure was adopted from that of Neddermann et al. (21). The slices were homogenized (using a microsample Teflon pestle (Scienceware) attached to a hand drill) until no large pieces of gel remained. The suspension was then vigorously shaken overnight at 37°C. The gel suspension was centrifuged at 14,000 rpm for 15 min at 4°C in the GS-3 rotor of a Sorvall RC-5B centrifuge. The supernatant was transferred to a 2-ml Eppendorf tube, and 1.6 ml of ice-cold acetone (4 volumes) were added. The samples were maintained in an ethanol/dry ice bath for 30 min and then centrifuged at 14,000 rpm for 10 min at 4°C. The supernatant was removed, and the pellet was washed with 600 l of 80% acetone, 20% dilution buffer (100 mM HEPES pH 8.3, 2 mM EDTA, 10 mM NaCl, 2 mM dithiothreitol, and 1 mg/ml BSA). The pellet was drained and dried at room temperature for 10 min followed by resuspension in 37.5 l of guanidine buffer (dilution buffer plus 6 M guanidine HCl). After incubation for 30 min at room temperature, 1.875 ml of dilution buffer was added, and the samples were incubated at 4°C overnight to permit renaturation.
The samples were concentrated in Centricon-10 microconcentrators that had been washed with 0.5 ml of dilution buffer containing 0.5 mg/ml BSA by centrifugation at 4°C at 6000 rpm for 45 min. The samples were then loaded onto the microconcentrators and centrifuged at 4°C at 6,000 rpm for 1 h. The resulting retentate (generally 75-125 l) was assayed for HMUDG activity.
Identification of HMUDG by Mass Spectrometry-Six gel slices containing proteins of apparent molecular mass between 23 and 35 kDa were excised from an unstained SDS-PAGE gel under a tissue culture hood to minimize contamination. Two-thirds of each gel slice was assayed for HMUDG activity as described above. The remaining third was fixed in 50% methanol, 5% acetic acid in water for 20 min and then washed with 50% methanol and water to remove the remaining SDS and acetic acid. The gel slices were cut into 1 mm 3 pieces, reduced, alkylated using iodoacetamide, and digested in-gel with trypsin (17). The peptides were extracted, dried under vacuum, and redissolved in 20 l of 5% acetic acid. Samples were desalted and concentrated using Zip-Tip TM C 18 micro-columns (Millipore). Peptides were eluted with 2 l of 70% acetonitrile. The final solution was adjusted to contain 50% acetonitrile, 1% acetic acid and then loaded into a gold-coated medium nanoelectrospray needle (Protana). MS and MS/MS spectra were acquired on a quadrupole time-of-flight tandem mass spectrometer equipped with a Z-spray nano-ESI ion source (Micromass, Manchester, UK). The instrument was calibrated in MS/MS mode using 1 pmol/l [Glu 1 ]-Fibrinopeptide B (Sigma) in 50% methanol/1% acetic acid. The capillary voltage used was 1400 V, and the cone voltage was set to 40 V. Tryptic peptide precursor ions (singly, doubly, or triply charged) were selected by the quadrupole mass filter (MS1) and induced to fragment by collision with Argon, and the resulting product ions were analyzed by the time-of-flight mass analyzer (MS2). The collision energy was varied between 15 and 45 V depending on the mass and charge-state of the peptides. The recorded MS/MS spectra were processed using MassLynx MaxEnt 3 (Micromass).
The charged state deconvoluted MS/MS spectra were directly used to search the public data base using the Mascot search program (Matrix Science, UK). For peptides with no exact matches in the public data bases, the MS/MS spectra were sequenced de novo either manually or with the aid of the PepSeq program (Micromass). The amino acid sequences obtained in this way were searched against the NCBI nonredundant protein data base using the BLAST search program (18).
Expression and Purification of hSMUG1-The hSMUG1 coding sequence was polymerase chain reaction-amplified as a BamHI-XhoI fragment from a hSMUG1 cDNA clone (ATCC 294606) purchased from ATCC. The polymerase chain reaction product was cloned into

FIG. 1. Purification and renaturation of HMUDG from SDS-PAGE.
A, gel electrophoresis of the most purified fraction of HMUDG. Indicated regions were cut out for analysis by mass spectroscopy and for assay of HMUDG activity. B, HMUDG activity of renatured protein. Renatured protein from the indicated regions of the gel was assayed as described under "Experimental Procedures." Assays were run for 12 h.
pGEX6P-1 (Amersham Pharmacia Biotech) so as to express the protein as an N-terminal GST fusion protein for purification on a glutathione column. Expression of the recombinant protein was induced in E. coli BL21-DE3 cells by the addition of isopropyl-␤-D thiogalactopyranoside to a final concentration of 0.5 mM. A 1-liter bacterial culture at an A 600 of 0.6 was induced at 30°C for 5 h and harvested by centrifugation at 4°C at 3200 ϫ g in a Beckman 6KR centrifuge. The recombinant protein was purified from the induced cell pellet using glutathione-Sepharose 4B affinity medium (Amersham Pharmacia Biotech) per the manufacturer's protocols for batch purification of fusion proteins. Following purification, the glutathione S-transferase moiety was removed according to the manufacturer's instructions using Prescission Protease (Amersham Pharmacia Biotech), leaving five amino acid residues remaining on the N terminus of the recombinant hSMUG1 (20).
Assay for Uracil DNA N-Glycosylase Activity by Cleavage of Ura Containing 2Ј-Deoxyribose Oligonucleotides-2Ј-Deoxyribose oligonucleotides were synthesized in the core facility of the NYU School of Medicine, Department of Cell Biology. All 2Ј-deoxyribose oligonucleo-  Assays were performed at 37°C in buffer containing 50 mM HEPES, pH 7.5, 20 mM NaCl, 0.1 mg/ml BSA, 1 mM EDTA, and 1 mM dithiothreitol. Enzyme, protein, and substrate were diluted to working conditions in assay buffer and equilibrated at 37°C. Assays contained 2 M 32 P-5Ј-end-labeled single-stranded or duplex 26-mer uracil-containing 2Ј-deoxyribose oligonucleotide and 20 nM hSMUG1 with or without 0. for 5 min, and products were separated by 20% PAGE in 7 M urea and Tris borate/EDTA (TBE). Products were analyzed via phosphorimaging using a Molecular Imager FX System with Quantity One Software (Bio-Rad) (20).

Purification of HMUDG from Calf Thymus-A mammalian
protein with HMUDG activity was purified from fresh calf thymus on the basis of the ability to excise [ 3 H]5hmUra from [ 3 H]SPO1 DNA, as shown in Table I. Calf thymus was used as the source of enzyme because large amounts of fresh tissue are available, and our previous studies revealed that the specific HMUDG activity of thymus was higher than any other tissue except brain (12). The purification scheme was similar to that used previously to purify human endonuclease III homologue 1 (15), with the addition of an (NH 4 ) 2 SO 4 concentration step. After the final purification step, the enzyme had been purified ϳ500-fold, with a yield of 0.2%. This estimate of degree of purity was based on the specific activity of the HMUDG in the crude preparation after the addition of 5 M NaCl and spooling out of the gelatinous nucleoprotein precipitate resulting from the addition of NaCl. This step actually increased the specific activity of HMUDG by 3-5-fold from that of the crude homogenate, so the final degree of purification (as measured by an increase in specific enzyme activity) was somewhere between 1500-and 2500-fold.
Polyacrylamide Gel Electrophoresis of Purified Protein and Renaturation of HMUDG-The most purified fraction (Fraction VIII) was analyzed by SDS-PAGE (Fig. 1A). The HMUDG protein could not be definitively associated with a single Coomassie Blue-stained band on the gel. To identify the region of the gel that contained HMUDG, an identical gel lane was cut into fractions (as indicated in Fig. 1A) and the proteins renatured. The renatured proteins were assayed for HMUDG activity. The greatest amount of HMUDG activity was present in slice 2 (Fig. 1B). The activity of renatured protein corresponded to a protein with an apparent molecular mass of 34 kDa, reasonably consistent with the molecular radius of 29 -30 kDa we observed for the native bovine protein on the Superdex-75 column (Fraction V).
Determination of Bovine HMUDG Amino Acid Sequence and Alignment with hSMUG1 Amino Acid Sequence-The nano-ESI mass spectrum (m/z range of 300 -1600) of peptides from slice 2 that showed the highest HMUDG activity is shown in Fig. 2A. In this complex spectrum, more than 150 singly, doubly, or triply charged tryptic peptides were observed, indicating that a mixture of proteins was present in this gel slice. In the MS survey, we found that the signals of a number of peptides observed in Fig. 2A were significantly weaker or absent in the nano-ESI spectra of other gel slices (data not shown), indicating that these peptides correlated with HMUDG activity. Therefore, these peptides were further fragmented by collisioninduced dissociation to generate sequences or sequence tags. The recorded MS/MS spectra were used for a data base search (see "Experimental Procedures"). To illustrate this identification process, the MS/MS spectrum of the doubly charged peptides at m/z 696.36 is shown in Fig. 2B. The complete y ion series (y 1 -y 14 ) could be assigned, allowing the determination of the full peptide sequence as SLAEGFLQELR (low collision energy MS/MS analysis as used here cannot distinguish between the isobaric Ile and Leu amino acid residues).
Direct data base searching with the uninterpreted MS/MS spectrum did not result in a positive protein identification. However, searching the NCBI protein data base using BLAST revealed that the bovine sequence nearly matched a peptide sequence from human SMUG1, a recently identified member of a family of uracil DNA N-glycosylases. Two changed amino acids were Ser3 Gly(G) and Glu3 Gln(Q). The probability that this identification was correct was high because a total of 13 peptide ions in Fig. 1A, corresponding to 11 peptide sequences obtained by MS/MS (Table II), were found to match or nearly match sequences from hSMUG1. The alignment of these bovine sequences with the hSMUG1 sequence is shown in Fig. 3.
Isolation of the cDNA of hSMUG1 and Expression of the Protein-The human SMUG1 cDNA was isolated and expressed as a GST fusion protein, which was purified and the GST moiety removed, leaving 5 amino acids on the N terminus of the recombinant hSMUG1. When analyzed by PAGE, the purified recombinant protein was visualized as a single band (Fig. 4) with an apparent molecular mass slightly larger than that of the region of the gel where the maximal activity of purified bovine HMUDG was recovered (Fig. 3).
Release of 5hmUra by hSMUG1 Compared with That of Purified Bovine HMUDG (Fraction VIII)-Purified expressed hSMUG1 was assayed for its ability to release [ 3 H]5hmUra from [ 3 H]SP01. At the concentrations assayed, there was a linear increase of release of 5hmUra with the increase in hSMUG1 concentration (Fig. 5). The identity of the released material was confirmed by HPLC. The specific activity of purified recombinant hSMUG1 was ϳ20ϫ greater than the most purified fraction (Fraction VIII) of bovine HMUDG.
Confirmation of Uracil DNA N-Glycosylase Activity of Recombinant hSMUG1-Recombinant hSMUG1 was analyzed for its ability to remove Ura from single-and double-stranded 2Ј-deoxyribose oligonucleotides containing a single Ura residue FIG. 6. Activity of hSMUG1 against uracil containing single-and doublestranded DNA containing matched and mismatched uracil. Recombinant hSMUG1 was assayed for the ability to cleave uracil containing 2Ј-deoxyribose oligonucleotides in the presence or absence of UGI (lanes 2-7). The oligonucleotides were either single-or double-stranded with uracil opposite A or G, as shown. The left lane shows oligonucleotide in the absence of enzyme. The right two lanes (8 and 9) show the activity of recombinant UDG in the presence or absence of UGI. (Fig. 6). The recombinant protein showed maximal activity in a cleavage assay against single-stranded DNA. hSMUG1 also displayed activity against double-stranded DNA containing a single Ura residue. As reported by Haushalter et al. (14), hSMUG1 activity was greater against mismatches (U:G) than against matches (U:A).
The activity of the recombinant hSMUG1 was distinct from UDG (in this case recombinant bacterial UDG). UDG had greater activity against double-stranded DNA containing U opposite A (Fig. 6, lane 7). Furthermore, UDG activity was suppressed by UGI in contrast to hSMUG1 activity, which was not (Fig. 6, lane 8). DISCUSSION SMUG1 was identified by binding to synthetic uracil DNA N-glycosylase inhibitors and has been definitively shown to be distinct from the primary uracil DNA N-glycosylase UDG and a second protein, thymine DNA N-glycosylase (TDG) which preferentially repairs U:G and T:G mismatches (14). In this study we have shown that a bovine protein with HMUDG activity, as defined previously by Hollstein et al. (8) and this group (9,12), has peptide sequences homologous to those of hSMUG1, whereas recombinant hSMUG1 is shown here for the first time to have HMUDG activity.
Previous observations of HMUDG activity in cells and tissues can now be reevaluated in light of the determination of the identity of HMUDG as SMUG1. The HMUDG activity was first searched for in bacteria, where it could not be found (10), and it was subsequently shown to be absent in yeast but present in more complex multicellular organisms (11). This unique phylogenetic distribution has been confirmed for SMUG1 (14) in contrast to UDG, which is highly conserved throughout phylogeny. The uracil DNA N-glycosylase activity of SMUG1 is not inhibited by the phage UGI. Our most purified fractions of bovine HMUDG also displayed uracil DNA N-glycosylase activity that was not inhibited by UGI (data not shown), further confirming the identity of HMUDG as SMUG1. Additional support for the identity of HMUDG as SMUG1 comes from the finding that SMUG1 ESTs are present in a wide variety of tissues (14,22), consistent with our previous observation that HMUDG activity is present in all tissues, and is most abundant in brain and thymus (11).
The biological role of HMUDG/SMUG1 is still uncertain. Haushalter and co-workers (14,22) speculated that SMUG1 might play a role in the repair of Ura residues resulting from deamination of Cyt residues in DNA that is transiently singlestranded during transcription, recombination, or replication. Recent work by the same group has suggested that SMUG1 functions as an antimutator, with preference for repair of Ura opposite guanine (22).
We previously proposed that HMUDG might repair 5hmUra residues arising from oxidation and deamination of 5meCyt residues in DNA that, if unrepaired, would be mutagenic. Such a hypothesis is supported by the common phylogenetic distributions of the use of 5meCyt as a regulatory element in transcription, HMUDG enzyme activity, and SMUG1 EST sequences. Recent evidence further supports this hypothesis. It was shown that 5hmUra, as part of 5hmUra:G mismatches, was repaired to a significantly greater extent than 5hmUra:A in identical sequences using partially purified HMUDG from Hela cells (23).
Regardless of which of the two putative functions is of greater significance to the maintenance of genomic integrity by HMUDG/SMUG1, it should be noted that a mutant Syrian hamster cell line (V79) deficient in HMUDG activity, which we developed, is viable and grows normally (24). Although it seems evident that HMUDG/SMUG1 protects cells from mutations arising from conversion of 5meCyt to 5hmUra and deamination of Cyt to Ura, the possibility remains that HMUDG/SMUG1 is also involved in a sequence of controlled oxidation of 5meCyt followed by deamination and removal of 5hmUra as a mechanism of changing the methylation status of specific regions of DNA.
Full understanding of the biological significance of HMUDG/ SMUG1 will require an explanation of both its unique in vitro substrate specificities and its phylogenetic distribution.