Biochemical characterization of uracil processing activities in the hyperthermophilic archaeon Pyrobaculum aerophilum.

: Deamination of cytosine to uracil and 5-methylcytosine to thymine represents a major mutagenic threat particularly at high temperatures. In double-stranded DNA, these spontaneous hydrolytic reactions give rise to G.U and G.T mispairs, respectively, that must be restored to G.C pairs prior to the next round of DNA replication; if left unrepaired, 50% of progeny DNA would acquire G.C –> A.T transition mutations. The genome of the hyperthermophilic archaeon Pyrobaculum aerophilum has been recently shown to encode a protein, Pa-MIG, a member of the endonuclease III family, capable of processing both G.U and G.T mispairs. We now show that this latter activity is undetectable in crude extracts of P. aerophilum. However, uracil residues in G.U mispairs, in A.U pairs, and in single-stranded DNA were efficiently removed in these extracts. These activities were assigned to a approximately 22-kDa polypeptide named Pa-UDG (P. aerophilum uracil-DNA glycosylase). The recombinant Pa-UDG protein is highly thermostable and displays a considerable degree of homology to the recently described uracil-DNA glycosylases from Archaeoglobus fulgidus and Thermotoga maritima. Interestingly, neither Pa-MIG nor Pa-UDG was inhibited by UGI, a generic inhibitor of the UNG family of uracil glycosylases. Yet a small fraction of the total uracil processing activity present in crude extracts of P. aerophilum was inhibited by this peptide. This implies that the hyperthermophilic archaeon ABSTRACT Deamination of cytosine to uracil and 5-methylcytosine to thymine represents a major mutagenic threat, particularly at high temperatures. In double-stranded DNA, these spontaneous hydrolytic reactions give rise to G•U and G•T mispairs, respectively, which must be restored to G•C pairs prior to the next round of DNA replication; left unrepaired, 50% of progeny DNA would acquire G•C → A•T transition mutations. The genome of the hyperthermophilic archaeon Pyrobaculum aerophilum was recently shown to encode a protein, Pa MIG, a member of the endonuclease III family, capable of processing both G•U and G•T mispairs. We now show that this latter activity is undetectable in crude extracts of P. aerophilum . However, uracil residues in G•U mispairs, A•U pairs and in single-stranded DNA are efficiently removed in these extracts. These activities could be assigned to a ~22 kDa polypeptide named Pa -UDG. The recombinant Pa -UDG protein is highly thermostable and displays a considerable degree of homology to the recently-described uracil DNA glycosylases from Archaeoglobus fulgidus and Thermotoga maritima . Interestingly, neither Pa-MIG , nor Pa -UDG are inhibited by UGI, a generic inhibitor of the UNG family of uracil glycosylases. Yet, a small fraction of the total uracil processing activity present in crude extracts of P. aerophilum could be inhibited by this peptide. This implies that this hyperthermophilic archaeon possesses at least a three-pronged defense against the mutagenic threat of hydrolytic deamination of cytosines in its genomic DNA.


INTRODUCTION
The integrity of genomic DNA of all organisms is under constant attrition by a variety of chemical and physical agents. A particular threat is posed by hydrolytic damage to DNA bases that carry exocyclic amino groups. The most frequent of these is the deamination of cytosine (1), which results in the formation of uracil, and therefore of G•U mismatches in doublestranded DNA. As uracil will base pair with adenine during the next round of DNA replication, cytosine deamination will result in a G•C → A•T transition mutation if unrepaired. Because cytosine hydrolysis is greatly enhanced at high temperatures (1), hyperthermophilic organisms might be expected to be more susceptible to this type of mutations. However, this does not appear to be the case, as the reported spontaneous mutation rates in thermophilic bacteria are similar to those observed in E. coli (2). Thus, hyperthermophilic organisms are apparently more efficient at repairing hydrolytic and oxidative damage to DNA bases (3). We decided to study how uracil is processed in the hyperthermophilic archaeon Pyrobaculum aerophilum (4), which has an optimal growth temperature of 100°C.
In most organisms, uracil is removed from DNA by uracil-DNA glycosylases (UDG) that are encoded by the UNG gene family (5). These highly-efficient enzymes remove the modified base from all substrates, be they G•U mismatches, A•U pairs or single stranded.
Interestingly, some organisms including Drosophila melanogaster appear to lack UNG gene homologues. Several years ago, we identified a novel mismatch-specific DNA-glycosylase, TDG, which removes uracil and thymine from G•U and G•T mispairs, respectively (6)(7)(8).
Although this enzyme is substantially less efficient than UDG in vitro, it appears to be widespread in nature, at least as far as its activity on G•U mispairs is concerned. E. coli contains the mug gene, which encodes a TDG homologue that can process G•U substrates, but not G•T mispairs (9). Eukaryotic TDG homologues are found also in Drosophila melanogaster and Schizosaccharomyces pombe (U. Hardeland, M. Bentele, J. Jiricny and P. Schär, manuscript in preparation). Due to their strong preference for G•U mispairs over A•U pairs or uracil in single-stranded DNA, these enzymes have been assigned primarily an antimutagenic role in DNA metabolism (9). Surprisingly, homology searches of the genome sequences available for several hyperthermophilic bacteria revealed neither UNG nor Mug homologues (10,11). However, Sandigursky and Franklin have recently identified uracil DNA-glycosylases encoded in the genomes of the thermophiles Thermotoga maritima (12) and Archaeoglobus fulgidus (13), which appear to be distantly related to the Mug class of enzymes. Interestingly, in contrast to the mismatch-specific Mug protein, these enzymes can remove uracil from DNA paudg4167as, 5' GTTGTATGAGGTACGGG 3' . A T7 promoter-specific primer was used for sequencing across the cloning junctions in the pET28c(+) vector.
P. aerophilum extracts and purified proteins P. aerophilum cultures were grown in the laboratory of Jeffrey Miller (University of California, Los Angeles, USA). The cell free extracts were prepared by Mahmud Shivji (ICRF, Clare Hall Laboratories, London, UK), according to the protocol described previously (16)  The PAE0651 open reading frame was amplified by PCR using the primers paudg3888s and paudg4434as containing NheI and BamHI restriction sites, respectively, and ligated into the corresponding restriction sites in the bacterial expression vector pET28c(+). The DNA sequence of the resulting plasmid (pET28-paudg) was confirmed by cycle sequencing using the primers paudg4183s, paudg 4167as and the T7 promoter primer. This cloning strategy places the PAE0651 coding sequence downstream of a vector-derived sequence encoding an N-terminal Histidine 6 -tag followed by a thrombin cleavage site. The N-terminal amino acid sequence of the resulting His 6 -Pa-UDG fusion protein is therefore MGSSHHHHHHSSG LVPRGS HMASDLQKL…. (bold residues indicate the authentic Pa-UDG N-terminus lacking the initiator methionine; the thrombin cleavage site is underlined).

Expression and Purification of Pa-UDG
The pET28-paudg expression construct was transformed into E. coli BL21(DE3) cells by electroporation. Following incubation at 30°C overnight on selective LB-plates supplemented with 2% D-glucose, single colonies were picked and used to inoculate 20 ml of LB-medium containing 50 µg/ml kanamycin and 2% D-glucose. Following an overnight incubation at 30°C, 20 ml of the culture were then used to inoculate 1 liter of LB-medium containing 50 µg/ml kanamycin, and the culture was grown at 30°C up to OD 600 of 0.7-0.9. The expression of Pa-UDG was then induced with 1mM isopropylthio-β-D-galactopyranoside (IPTG, Sigma).
After 5 h of incubation at 30°C, the cells were spun down at 8000 rpm and 4°C for 10 min in a Sorvall SLA-3000 rotor.
The cell pellet was resuspended in 3 ml/mg sonication buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 10% glycerol, 1 mM imidazole, 0.25% Tween-20, 10 mM β-mercaptoethanol and 1 mM phenylmethylsulfonyl fluoride). The cells were lysed by sonication on ice (25 × 5 s bursts with intermittent chilling for 10 s). The sonicate was clarified by centrifugation at 15000 rpm at 4°C for 30 min in a Sorvall SS34 rotor. The supernatant (soluble fraction) was removed and incubated with gentle shaking for 1 h at 4°C with 2 ml Ni-NTA agarose (QIAGEN), pre-equilibrated in sonication buffer. The suspension was then packed into a disposable column, from which the unbound proteins were eluted with sonication buffer containing increasing concentrations of imidazole (1 × 10 column-volumes (cv) 5 mM imidazole, 2 × 2.5 cv 60 mM imidazole). The histidine-tagged Pa-UDG protein was eluted with 3 × 1 cv of sonication buffer containing 250 mM imidazole. The 250 mM fractions were pooled and dialyzed overnight at 4°C against 2 liters of binding buffer (50 mM sodium phosphate, pH 8.0, 10 mM NaCl, 10 % glycerol, 10 mM β-mercaptoethanol). The dialyzed fraction was loaded onto a 1-ml HiTrap SP FPLC column (Amersham Pharmacia Biotech) and the column was washed with 10 ml of binding buffer. The column was then eluted with a 20 ml linear gradient of 10-600 mM NaCl. The nearly-homogenous Pa-UDG protein eluted as a major peak in fractions containing 150-200 mM NaCl. These fractions were pooled and dialyzed against storage buffer (50 mM sodium phosphate, pH 8.0, 50 mM NaCl, 10 % glycerol, 10 mM βmercaptoethanol). The concentration was estimated to be ~1 mg/ml, as determined by the method of Bradford (18). The protein was stored in small aliquots at -80°C. This preparation was assayed for UDG activity.

Glycosylase Assays
The uracil glycosylase activity of P. aerophilum extracts or of the purified enzymes was monitored by means of a standardized "nicking assay", as described by Hudepohl et al. (16).
In this assay, removal of uracil or thymine from a fluorescently-labeled synthetic 60-mer oligonucleotide DNA substrate generates an abasic site, which can be subsequently cleaved by

Uracil-processing activities in P. aerophilum extracts
Spontaneous hydrolytic deamination of cytosines in DNA is greatly enhanced at high temperatures (1) and generates uracil bases mispaired with guanines. Since the optimal growth temperature of P. aerophilum is 100˚C, but its spontaneous mutation rate is no higher that that of E coli (2), we anticipated that this organism should possess an efficient base excision repair mechanism, capable of rapid restoration of the original G•C base pairs. We first decided to examine the removal of uracil from DNA in crude extracts of P. aerophilum. This was measured as the efficiency of uracil-directed DNA strand incision upon incubation of different DNA substrates with the protein extract. As shown in Figure 1 coli Mug (9), H. sapiens SMUG1 (21) and H. sapiens MBD4 (22). This suggested that uracil processing in the P. aerophilum crude extract could be accounted for by at least two distinct type of DNA glycosylase, a major activity that is not inhibited by UGI, and a minor activity that is sensitive to this peptide. However, homology searches of the P. aerophilum genome failed to detect genes encoding any UNG-type proteins.
In order to test the effect of temperature on uracil removal from either the doublestranded G•U substrate or from single-stranded DNA, the nicking assay was repeated at different incubation temperatures. The uracil-processing activities present in the P. aerophilum extracts were stable at temperatures up to 90˚C (Fig. 2), which was anticipated for enzymes of a hyperthermophile. Interestingly, while the processing efficiency of the single-stranded substrate was largely unaffected by incubation temperature, the 60-mer G•U substrate was processed substantially more efficiently at 70˚C than at lower temperatures. This suggests that the enzymatic activity increases with temperature, with the optimum lying between 70°C and 90°C under our assay conditions. The activity increase seen at 70˚C was not due to melting of the oligonucleotide duplex; first, this would imply a further improvement at 90˚C, and second, the single-stranded substrate was processed less efficiently at 70˚C than the G•U heteroduplex.
However, melting was probably responsible for the observed activity decrease at 90˚C (Fig. 2, lane 4), as the processing of the G•U substrate decreased to the level seen with the singlestranded oligonucleotide (Fig. 2, lane 8).

G•T mismatch processing in P. aerophilum extracts
At 100˚C, deamination of 5-methylcytosines is accelerated in parallel with that of cytosines (1).
While the latter process generates G•U mispairs in double-stranded DNA, the former reaction gives rise to G•T mismatches. Several years ago, we described a mismatch-specific thymine/uracil DNA-glycosylase, TDG, which removes thymine and uracil from G•T and G•U mismatches (8). We were subsequently able to identify and characterize a bacterial homologue of TDG, named Mug, which removes uracil from G•U mispairs but which has no G•T processing ability (9). At about the same time, Fritz and colleagues identified a similar activity in the thermophile M. thermoautotrphicum, named Mth.MIG (23), which also processes G•U and at Hauptbibliothek Universitaet Zuerich Irchel. Bereich Forschung on March 10, 2014 http://www.jbc.org/ Downloaded from G•T mispairs in a manner analogous to TDG, but which belongs to the endonuclease III family of DNA glycosylases. Notably, the MIG gene homologue was recently identified also in P.
aerophilum by Miller and colleagues and the recombinant protein expressed in E. coli was indeed shown to be a mismatch-specific thymine/uracil DNA-glycosylase (14). We decided to test whether the Pa-MIG activity could be detected in the P. aerophilum crude extracts. Figure 3 (lanes 4 and 6) documents that purified recombinant Pa-MIG enzyme removes uracil and thymine mismatched with guanine with about equal efficiency. In contrast, no G•T processing activity could be detected in the crude cell extracts (Fig. 3, lane 5), even when the sensitivity of the assay was increased by the use of 32 P-labelled 90-mer substrates (data not shown). These results, together with the fact that Pa-MIG enzyme does not remove uracil residues from A•U pairs or from single stranded DNA, imply that this enzyme does not significantly contribute towards the efficient uracil-processing activity observed in the P. aerophilum extracts.

Identification of a P. aerophilum uracil-DNA glycosylase (Pa-UDG)
During the course of these biochemical studies, the complete sequence of the P. aerophilum genome became available (24). We searched this genome database for open reading frames encoding homologues of the UNG or the Mug types of uracil DNA glycosylases, but failed to find likely candidates. This situation changed following the identification of a uracil-DNA glycosylase in the thermophilic eubacterium T. maritima, Tm-UDG (12). This enzyme was shown to be distantly related to Mug (9), but displayed a substrate specificity similar to that detected in the P. aerophilum extracts. We therefore carried out a homology search using the amino acid sequence of Tm-UDG. This yielded an ORF encoding a polypeptide with significant homology to Tm-UDG (12) and Af-UDG (13) (Fig. 4A). A putative translation start was assigned from the genomic sequence and appears to be a GUG rather than an AUG. This is, however, plausible, because GUG is known to be used frequently (up to 25%) as a start codon of archaeal genes (25). The resulting 196-amino acid protein has a calculated molecular mass of 21563 Daltons and a theoretical pI of 9.39. A clustered relationship analysis with UNG, TDG, SMUG and thermophilic uracil DNA glycosylases revealed the existence of four distinct groups of closely-related proteins, where Pa-UDG falls into the category of the thermophilic enzymes ( Fig. 4B).

Molecular cloning and expression of the P. aerophilum UDG coding sequence; purification of recombinant Pa-UDG enzyme
The putative paudg sequence (PAE0651) was amplified from a genomic clone by high fidelity PCR. The PCR primers used contained restriction sites for NheI and BamHI, which were then utilized in cloning of the amplified fragment into the corresponding sites within the polylinker of the expression vector pET28c(+). The putative start codon GTG was omitted in the upstream primer, because a suitable ATG translation start site was provided in the expression vector sequence. The amino terminal of the recombinant protein thus acquired a 6-histidine-tag, as well as a thrombin cleavage site, should the tag need to be removed at a later stage (see Experimental Procedures). The pET28c(+)-paudg construct was transformed into competent E. coli BL21(DE3) cells and the expression of paudg was induced with IPTG. Crude lysates of the IPTG-induced cells contained a protein migrating with an apparent molecular size of ~25 kDa in SDS-PAGE, which was absent from extracts of uninduced cells (data not shown). This is consistent with the predicted MW of a recombinant His 6 -tagged Pa-UDG of 24 kDa. The overexpressed protein was purified from the E. coli extract using a two-step fractionation scheme involving metal affinity chromatography (nickel-nitrilotriacetic acid) and FPLC (HiTrap SP) (Fig. 5). Starting from one liter of induced E. coli culture, we typically obtained ~4 mg of nearly homogeneous recombinant Pa-UDG protein at a concentration of ~1 mg/ml. The fraction shown in lane 3 of Figure 5 was used in the subsequent biochemical assays.

Pa-UDG is a thermostable, monofunctional uracil DNA-glycosylase that is resistant to UGI
Monofunctional DNA glycosylases generate apurinic or apyrimidinic (AP) sites in DNA, but are unable to convert these into strand breaks, as they lack an associated DNA-lyase activity. Thus, under the conditions of the in vitro nicking assay used in this study, the oligonucleotide substrates had to be cleaved at the resulting AP-sites either by the addition of an APendonuclease such as HAP1 (26), or by hot alkaline treatment. In contrast, bifunctional DNA glycosylase/lyases can carry out both the base removal and sugar-phosphate backbone cleavage reactions (27). It is impossible to differentiate between these two enzyme classes in assays using crude cell extracts, because these contain also AP-endonuclease activities that can cleave the substrate DNA at the AP-sites. In order to be able to assign Pa-UDG to its appropriate class, we carried out the in vitro nicking reaction with the purified recombinant enzyme either in the presence or in the absence of HAP1 or NaOH. As shown in Figure 6A, incubation of the 60mer G•U substrate with the enzyme gave rise to only a small amount of the 23-mer product (lane 4), which arose as a result of spontaneous β-elimination at the AP-site under the conditions of the assay. In contrast, when the oligonucleotide duplexes were subjected to HAP1 or hot alkaline treatment after the incubation with Pa-UDG, most of the 60-mer substrate was converted to a 23-mer product, having either a hyroxy group at the 3'end in the case of HAP1 digestion (lane 5) or a phosphate group due to a β,δ-elimination reaction in the case of NaOH treatment (lane 6). This showed that Pa-UDG generates AP-sites in the substrate DNA, which can be cleaved in a subsequent reaction with HAP1 or NaOH. Pa-UDG is thus a monofunctional uracil DNA-glycosylase. The recombinant protein did not lose activity when pre-incubated without substrate at 85°C for 15 minutes. This eliminates the possibility that the uracil processing observed in these assays was due to a contamination with UDG from E. coli, since the latter activity was reduced to undetectable levels following such heat treatment (data We further examined whether the recombinant Pa-UDG is inhibited by UGI. Figure 6B shows that G•U processing by Pa-UDG is not affected by the presence of up to 10-fold excess of the inhibitor peptide in the reaction, while an equivalent activity provided by the E. coli Ung protein was completely inhibited under similar conditions. We therefore conclude that Pa-UDG cannot account for the minor UGI-sensitive uracil DNA glycosylase activity discernible in the P. aerophilum extracts.

Substrate specificity of the purified recombinant Pa-UDG
We next decided to test whether the substrate specificity of the purified recombinant Pa-UDG enzyme corresponded to that observed in the crude P. aerophilum extracts. Figure 7A shows that this is indeed the case: the enzyme removes uracil from G•U, A•U and single-stranded DNA substrates with an order of preference comparable to that seen in the crude extracts (Fig.   1B). Kinetic time course experiments, in which 1 pmol of G•U, ssU or A•U substrates was incubated with 100 fmol of Pa-UDG at 70°C clearly established that the enzyme acts more efficiently on the G•U substrate that on uracil-containing single-stranded DNA or on the A•U duplex (Fig. 7B). 80% of the G•U substrate was cleaved after 30 minutes of incubation, indicating an at least 8-fold enzymatic turnover in this reaction. The activity on the ssU and A•U substrates was lower, with only ~30% being processed after 30 minutes under similar assay conditions. No activity was discernible with the G•T substrate (not shown). The observed turnover on the G•U mispair is in line with the biochemical properties of UNG-type uracil DNA glycosylases, but contrasts with those of TDG and Mug enzymes that were shown to be strongly product-inhibited (9,28,29). Correspondingly, the latter enzymes bind to oligonucleotide substrates containing the product AP-site in electrophoretic mobility shift assays (EMSA) with high efficiency. Pa-UDG failed to form a protein/DNA complex detectable in EMSA with any of the substrates tested (data not shown).
Taken together, the resistance of purified recombinant Pa-UDG to the UGI peptide, its ability to turn over in in vitro assays and its preference for G•U substrate over A•U and uracil in single-stranded DNA leads to the conclusion that this enzyme is different from UNG or Mug type uracil DNA glycosylases and that is most likely responsible for the major uracil processing activity observed in the crude P. aerophilum extracts.

DISCUSSION
The extracts of P. aerophilum contain an efficient uracil-processing activity. We identified and expressed in E. coli a P. aerophilum ORF that encodes a putative homologue of thermophilic uracil DNA glycosylases, Pa-UDG. We showed that the recombinant protein acts as a monofunctional uracil DNA-glycosylase that can remove the aberrant base from single stranded DNA, as well as from A•U pairs and G•U mispairs. As the substrate specificity of this enzyme corresponded to that seen in the total extracts of P. aerophilum, we propose that this enzyme accounts for the major uracil processing activity in the extracts of the hyperthermophilic archaeon. In contrast, the Pa-MIG activity, which removes thymine and uracil residues from mispairs with guanine (14) could not be detected in the extracts of this organism.
Pa-UDG is most closely related to two recently identified thermostable uracil DNA glycosylases from T. maritima and A. fulgidus (12,13), and these three proteins branch off as a separate cluster of uracil DNA glycosylases in phylogenetic analysis (Fig. 4B). This distinction is not only evident at the level of sequence similarity, but also by their biochemical properties such as substrate specificity, stability of protein/DNA interactions and enzymatic turnover.
However, the amino acid sequences around the putative active sites of the different uracil glycosylases are interesting and may help us understand more about the evolution and function of these proteins (see also references (30,31)). Crystal structures of the UNG-type uracil glycosylases from Herpes simplex virus (32) and Homo sapiens (33), as well as that of the E.
coli Mug protein (34,35), have shown that the enzymes have a structurally conserved, bipartite active site (Fig. 8). Motif A in the UNG-type enzymes contains the sequence GQDPYH…F, where the aspartic acid residue (shown bold in a gray field) activates a molecule of water towards a nucleophilic attack on the C1' of the sugar residue bound to the aberrant base. The tyrosine ensures exclusion of thymine from the active site pocket, while the phenylalanine in the second conserved motif stabilizes uracil in the pocket via π−π interactions. The histidine residue of the conserved B-motif HPSPLS is thought to activate the glycosidic bond towards nucleophilic attack by the water molecule through binding to the O 2 of the uracil (32). In the Mug structure (34), both the aspartate and the histidine are replaced with asparagines, which can still position the water molecule correctly for the attack, but cannot activate it as efficiently. This was thought to be the underlying reason for the low enzymatic activity of Mug and TDG as compared to the UNG-type UDG enzymes (34,35). However, the issue must be somewhat concerned. However, unlike the Mug family, the A motif of the thermophiles has an acidic residue (E), which may well assist in the activation of the catalytic water molecule, as suggested for the UNG-type enzymes (30,31). The tyrosine residue in the GQDPY motif of the UNGtype UDGs was proposed to prevent the binding of thymine in the active site pocket (32,33) and it may be assumed that substitution of this residue for the small glycine would allow the processing of this base. While this may be true for UNG-type UDG's (36), of all the enzymes with the GxxPG motif (Fig. 8) only the human TDG removes thymines from G•T mispairs (9), which suggests that the thymine exclusion criteria are more complex than previously thought.
The A-motif of the hyperthermophiles also contains several acidic residues (shown in bold in Analysis of the B-motif is also interesting. The thermophilic UDGs, as well as the UNG-and SMUG-type enzymes, share the conserved sequence HPS (or HPA in the archaea), while the histidine residue is absent from the corresponding site of the Mug enzymes (Fig. 8).
Comparison of the substrate specificities of the HPS/A group revealed that they are all able to process single-stranded DNA substrates, while Mug and TDG cannot. However, the Mug homologue from S. pombe has the sequence GIS, thus lacking both the histidine and the proline residues, yet it can remove uracil from ssDNA (U. Hardeland and P. Schär, unpublished). It thus appears unlikely that the histidine in this motif is required for ssDNA processing.
Interestingly, following removal of the uracil, the Mug enzymes remain bound to the AP-site containing oligonucleotides (9,29) and hence fail to turn over. In contrast, all the HPS/A enzymes, with the notable exception of SMUG, display turnover kinetics. The crystal structure of substrate-bound Mug revealed that, similarly to the UNG-type proteins, this enzyme inserts a wedge of three amino acids into the DNA duplex in order to fill the space vacated by the flippedout uracil, thus preventing the neighboring bases from collapsing onto each other. The structural organization of this wedge involves the residues of Mug motif B. Unlike UNG, however, Mug establishes specific hydrogen bonds with the Watson-Crick face of the widowed guanine opposite the abasic site, as well as with the sugar residues of the complementary strand (34).
Since the UNG-(32,33) and the thermophilic-type UDG's do not appear to engage in rigid complementary strand interactions, it may be this property of Mug-type enzymes and thus this function of motif B that is responsible for the observed differences in turnover kinetics. This hypothesis would appear to be supported by the finding that a MPS to HPS mutation within the motif B of the human TDG generated an active enzyme that has lost its ability to stably interact with the DNA substrate (29). However, though it may be an important contributing factor, it is unlikely that a single amino acid sequence motif is entirely responsible for controlling the dissociation of the enzyme from its substrate and further studies are required in order to elucidate these differences.
The genome of P. aerophilum (24) contains two ORFs encoding conserved uracil glycosylases: Pa-MIG (14) and Pa-UDG. Both these enzymes could be expressed in E. coli and subjected to biochemical analysis, which revealed that they were not inhibited by the generic uracil glycosylase inhibitor UGI. As similar experiments carried out with crude extracts of P. aerophilum revealed a low degree of inhibition of uracil processing, we conclude that they must contain yet another enzyme, capable of addressing this pre-mutagenic product of hydrolytic deamination of cytosine. Homology sequence searches failed to find a third ORF that would contain the two amino acid motifs thought to comprise the active site of the UDG family of enzymes. We are currently attempting to identify this elusive activity by biochemical means.         Shown are partial amino acid sequence alignments of the active site motifs A and B (see also