Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

The isolation of two functional NAT* loci in humans a decade ago spurred a litany of molecular genetic and biochemical studies on NAT polymorphisms. The two loci, NAT1* and NAT2*, are independently regulated and encode proteins with 81% deduced amino acid sequence similarity. Both NAT1 and NAT2 catalyze N-acetylation (detoxification) and O-acetylation (activation) reactions, and possess selectivity with various therapeutic agents and with carcinogenic and mutagenic compounds. In N-acetylation reactions, NAT1 has specificity for the drugs 4(p)-aminosalicylate and 4-aminobenzoate and selectivity for several carbocyclic arylamine carginogens occurring as environmental/occupational pollutants (e.g., 2-aminofluorene, 4-aminobiphenyl, and benzidine), whereas NAT2 has a preference for sulfamethazine. For O-acetylation and N,O-transacetylation reactions, NAT1 is selective with carbocyclic arylamines and NAT2 with dietary heterocyclic amine mutagens. Overall, the data indicate that NAT1 is a much more efficient and effective catalyst than NAT2 (cf. Vatsis et al., 1995; Grant et al., 1997; Vatsis and Weber, 1997).

The initial focus on NAT2* soon generated unequivocal evidence for association of this locus with the human genetic variability that had been amply documented with isoniazid and sulfamethazine, and typically illustrated as a bimodality, and often trimodality, in the distribution of populations for N-acetylation of these compounds. The pronounced variation and ethnic distribution of NAT2* alleles have also been thoroughly investigated, and, in general, NAT2* genotypic-NAT2 phenotypic indices are reasonably well correlated. Of the 26 human NAT2* allelic variants known to date, five to six remarkably account for almost all of the variation seen in diverse ethnogeographic populations [reviewed in Vatsis and Weber (1997) and in Grant et al. (1997)].

Although a wealth of publications had shown appreciable differences in N-acetylation of 4-aminobenzoate and 4-aminosalicylate [up to 90-fold in vitro and more than 100-fold in vivo (cf. Vatsis and Weber, 1994)], the human acetylase with activity toward these compounds was long regarded as genetically invariant (Weber, 1987) because the existence of two independently regulated NATs with discrete substrate selectivities had not yet been discovered. Strong indirect evidence for an acetylation polymorphism at the human NAT1* locus was provided by disclosure of NAT1* structural heterogeneity (Vatsis and Weber, 1993), and demonstration of a tendency toward bimodality in the distribution in vivo (Grant et al., 1992) and in vitro (Weber and Vatsis, 1993) of established NAT1-specific substrates (Grant et al., 1991). The evidence was indirect in that NAT1* allelic variation was observed with persons of unknown NAT1 phenotype (Vatsis and Weber, 1993; Vatsis et al., 1994), and, conversely, NAT1 phenotypic variation was determined with subjects whose NAT1* genotype had not been ascertained (Grant et al., 1992; Weber and Vatsis, 1993; Vatsis and Weber, 1994). These concerns have begun to be addressed, as evidenced by three very recent studies on direct phenotypic-genotypic correlations: elimination of 4-aminosalicylate in vivo and N-acetylation by whole blood and cytosol of heterologously expressed alleles from individuals of defined NAT1* genotype (Hughes et al., 1998); 4-aminosalicylate N-acetylation by several structurally characterized recombinant NAT1 variants (Lin et al., 1998); and 4-aminobenzoate N-acetylation by whole blood from 85 unrelated subjects of established NAT1* genotype, with 8% of the bimodally distributed population displaying the slow NAT1 phenotype (Butcher et al., 1998).

In the present study, we have found a prevalence of wild-type NAT1*4 and mutant NAT1*10, and a considerably lower distribution frequency for mutant NAT1*11 and NAT1*16 alleles, identified by direct sequencing of 1.6-kb fragments of PCR-generated NAT1* from an interethnic group of 38 unrelated subjects. Heterologous expression in bacterial and mammalian cells and rigorous biochemical and immunochemical characterization of the recombinant alleles, as well as of cytosol from mononuclear leukocytes of NAT1*4/*4 and NAT1*10/*10 homozygotes, revealed no major dissimilarities between wild-type NAT1*4 and mutants NAT1*10 (polyadenylation signal mutation) and NAT1*11 (coding and 3' noncoding exon mutations). In NAT1*16, an AAA insertion on the 3' side of the AATAAA signal resulted in a 2-fold decrease in immunoreactivity and a commensurate decline in N-acetylation activity of NAT1 16. This finding constitutes the first example of a causative relationship between NAT1 phenotype and a mutationally modified sequence in the 3' untranslated region of an NAT* variant.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials

Restriction endonucleases and other DNA-modifying enzymes were obtained from GibcoBRL/Life Technologies (Bethesda, MD), New England Biolabs (Beverly, MA), Promega (Madison, WI), and Boehringer Mannheim (Indianapolis, IN). Wizard PCR Preps DNA purification system was obtained from Promega, SeaPlaque GTG agarose from FMC Corporation (Rockland, ME), [-³²P]dATP and the Sequenase version II kit from Amersham (Arlington Heights, IL), and Immobilon-P transfer membranes from Millipore (Bedford, MA). USB Specialty Biochemicals was the source of inorganic pyrophosphatase, and of 7-deaza-dGTP and dITP termination nucleotide mixes. The pET expression system, which includes the pET-17b vector and Escherichia coli strains Novablue and BL21(DE3)pLysS, was purchased from Novagen (Madison, WI). Prokaryotic and eukaryotic cell culture media were from Difco (Detroit, MI) and GibcoBRL, respectively, the eukaryotic TA cloning kit from Invitrogen (San Diego, CA), fetal bovine serum from Hyclone (Logan, UT), the -galactosidase enzyme assay system from Promega, ¹²⁵I-conjugated goat anti-rabbit secondary antibody from DuPont NEN (Boston, MA), Biomax MR film from Eastman Kodak Company (Rochester, NY), and the protein assay and silver stain kits from Bio-Rad (Hercules, CA). Antibodies and molecular biology grade chemicals and buffers for recombinant DNA experiments were from Sigma (St. Louis, MO). Oligonucleotide primers were made at the University of Michigan DNA Synthesis Facility and purified by HPLC. The pGEX-KN vector (Hakes and Dixon, 1992) and E. coli strain BL21 were gifts from Dr. Jack E. Dixon of the Department of Biological Chemistry at this institution. E. coli strain JM105 harboring the plasmid pKK2233-NAT2*4 was kindly donated by Dr. David W. Hein, Department of Pharmacology and Toxicology, University of Louisville School of Medicine.

Vacutainer tubes were purchased from Becton-Dickinson (Rutherford, NJ). The red blood cells lysis buffer and phosphate-buffered saline, as well as the Polymorphprep density gradient medium, a sterile solution of 13.8% (w/v) sodium metrizoate and 8.0% (w/v) dextran 500, were obtained from GibcoBRL.

Leupeptin, phenylmethylsulfonylfluoride, acetyl-DL-carnitine, carnitine acetyltransferase, and acetyl CoA were purchased from Sigma. Arylamine substrates, two arylamide derivatives [phenacetin (p-ethoxyacetanilide) and 2-acetamidofluorene], and (+)amethopterin were from Aldrich (Milwaukee, WI). The acetylated derivative of p-anisidine (p-methoxyacetanilide) was obtained from Pfaltz and Bauer (Waterbury, CT), 4-acetamidobenzoate was from Eastman Kodak, and 5-acetamidosalicylate was a generous gift from Dr. Marshall Montrose (The Johns Hopkins University School of Medicine, Baltimore, MD). Acetylated 4-aminosalicylate was synthesized by one of us (K. P. Vatsis) with the help of Dr. Alfin D. N. Vaz of the Department of Biological Chemistry at this institution, and the purity of the compound was ascertained by gas chromatography-mass spectrometry and HPLC.

Subjects

A total of 26 healthy and unrelated volunteers (19 males; 7 females) were recruited from the graduate student body, postdoctoral trainees, faculty, and staff at the University of Michigan Medical School. To avoid genetic "ad-mixtures" as much as possible (Lin et al., 1993), the ethnically diverse population studied consisted of individuals born in Japan (n = 5), Korea (n = 5), mainland China (n = 6), India (n = 5), and Latin America (Mexico and South America; n = 5). Written consent was obtained from all volunteers, and the experimental procedures were approved by the University Human Subjects Committee.

Blood samples (30-40 ml) were collected in yellow-capped vacutainer tubes at the outpatient clinics of the University of Michigan Hospital. The blood was submitted to density gradient centrifugation for isolation of mononuclear and polymorphonuclear leukocytes, from which genomic DNA and the cytosolic fraction were prepared for NAT1* amplification by PCR and for catalytic activity determinations, respectively.

Methods

Amplification and Direct Sequencing of NAT1*. The separation of mononuclear and polymorphonuclear leukocytes from whole human blood is outlined toward the end of the Experimental Procedures section. Separated white cells corresponding approximately to 40% of the blood sample were mixed well with 4 volumes of phosphate-buffered saline, and centrifuged at 12,000g for 15 min (4°C) in a fixed angle rotor. The pellets were resuspended in an equal volume of 10 mM Tris-HCl buffer (pH 7.5) containing 1 mM EDTA, and combined. Genomic DNA template was extracted from the pooled white blood cells (Vatsis and Weber, 1993), and the size and intactness of the preparation verified by electrophoresis on 1% agarose gels.

Heterologous Expression of Human NAT1* Alleles: Transformation of E. coli. pCR3-NAT1*4 and pCR3-NAT1*11 recombinant plasmids (see COS-1 cells below) were digested with BamHI and EcoRI to release the NAT1*4 and *11 inserts, which were ligated with similarly cleaved and dephosphorylated pET-17b prokaryotic expression vector with a T7 promoter, and electroporated into E. coli cloning host Novablue. The recombinant plasmids pET-17b-NAT1*4 and pET-17b-NAT1*11 were mapped with SspI to confirm NAT1* insert orientation, and sequenced to establish unambiguously the identity of the NAT1* alleles. The T7 tag of the pET-17b vector was excised by digestion with NdeI and BamHI, blunt ends were generated by treatment with Mung bean nuclease, and the plasmids were electroporated into E. coli expression host BL21(DE3)pLysS as described in the Bio-Rad instruction manual.

Heterologous Expression of Human NAT1* Alleles: Transient Transfection of COS-1 Cells. The coding exon and 3' UTR of NAT1* (nt 1-1175) were amplified with template DNA from NAT1*4/*11 and NAT1*10/*16 heterozygotes; the sense and antisense amplification primers contained BamHI and EcoRI restriction sites, respectively, to facilitate transfer of NAT1* inserts to other vectors. The PCR products were ligated directly into eukaryotic expression vector pCR3 with the TA cloning system, yielding recombinant plasmids pCR3-NAT1*4, pCR3-NAT1*10, pCR3-NAT1*11, and pCR3-NAT*16, which were screened for the presence of insert by the procedure of Akada (1994). Positive colonies on 0.8% agarose gels were identified by size, insert orientation in positive clones was verified by mapping with BsaI, and the identity of the NAT1* alleles was conclusively affirmed by sequencing of the 1.2-kb fragments. Since each NAT1* insert contained the 3' UTR with its own polyadenylation signal, the bovine growth hormone polyadenylation signal of the pCR3 vector was removed by cleavage with NotI and AflIII. Blunt ends were generated with T4 DNA polymerase, the recombinant vectors were submitted to electrophoresis on 1% SeaPlaque GTG agarose gels, and were then ligated and electroporated in E. coli cloning host Top10 F as described in the instruction booklet from FMC Corporation. The reporter vector pCR3Gal was constructed by ligation of the E. coli -galactosidase gene with the NotI restriction site in the polylinker region of previously dephosphorylated pCR3 vector, and was mapped with EcoRV to confirm orientation. All recombinant plasmids were purified by double-banding CsCl gradient centrifugation (Sambrook et al., 1989).

pCR3Gal

Preparation of Polyclonal Antibody to Human NAT1. PCR-generated NAT1*4 (nt 1-1175) with NotI and EcoRI restriction sites at the 5' and 3' ends, respectively, was extracted with phenol/chloroform, digested with NotI and EcoRI for 20 h, and purified by agarose gel electrophoresis in conjunction with the Wizard PCR Preps DNA Purification System. Digested and gel-purified NAT1*4 was ligated in the NotI and EcoRI sites of previously dephosphorylated pGEX-KN prokaryotic expression vector, which contains the glutathione S-transferase gene from Schistosoma japonicum as a fusion partner. The recombinant plasmid (pGEX-KN-NAT1*4) was mapped with PstI to confirm insert orientation, and the identity of NAT1*4 was established unequivocally by sequencing.

Preadsorption of anti-NAT1 Serum to E. coli Lysates. Nonspecific immunoreactive components were eliminated by adsorption of human NAT1 antiserum to bacterial lysates before immunoblotting. To this end, E. coli strain BL21(DE3)pLysS bearing expression vector pET-17b without NAT insert was grown overnight in Luria-Bertani medium in the presence of ampicillin (100 µg/ml). The culture was centrifuged at 4,000g for 15 min (4°C), and pellets were resuspended in 5 mM Tris-HCl buffer (pH 7.5 at 37°C) containing 3 mM EDTA, 2 mM dithiothreitol, 0.02 mM leupeptin, and 0.1 mM phenylmethylsulfonylfluoride. Native and heat-denatured E. coli suspensions (1 mg of protein/ml) were added to a solution of 0.2% Tween 20 in 0.15 M NaCl-0.02 M Tris-HCl buffer (pH 7.4) containing appropriately diluted antiserum, the mixtures were rotated overnight at room temperature, centrifuged at 8,000g for 10 min (4°C), and the supernatant fraction submitted to immunoblotting.

Immunoblotting. Cytosol from transformed bacteria and transfected COS-1 cells was submitted to electrophoresis on 12% polyacrylamide gels in the presence of sodium dodecyl sulfate, and electrophoretically transferred to Immobilon-P membranes. The membranes were incubated for 1 h at room temperature in 0.15 M NaCl-0.02 M Tris-HCl buffer (pH 7.4) containing 2% bovine serum albumin and 1 mM EDTA, the solution was supplemented with appropriately diluted NAT1 antiserum that had been previously adsorbed, and the incubation was continued overnight at 4°C. After washing with 0.15 M NaCl-0.02 M Tris-HCl buffer (pH 7.4), the filters were incubated for 1 h with goat anti-rabbit ¹²⁵I-conjugated secondary antibody, and exposed to film for 18 to 20 h at 70°C. Quantitation of NAT1 bands on autoradiograms was achieved by densitometric scanning (Bio-Rad model GS-670 Imaging Densitometer and accompanying software). Without exception, the amount of protein expressed was obtained from the slope of the line for the relative band density (in arbitrary units) plotted against different amounts of cytosolic protein to ensure the direct proportionality of these two parameters.

Isolation of Human White Blood Cell Types and Preparation of Cytosol. Anticoagulated whole human blood (3-5 ml) was layered at room temperature over an equal volume of sodium metrizoate-dextran density gradient medium (1:1.7; Polymorphprep solution). The samples were centrifuged at 500g for 20 to 30 min (20°C) in the swinging bucket rotor of a tabletop centrifuge, resulting in separation of the leukocytes into two clearly demarcated bands: an upper band consisting of mononuclear leukocytes and a lower band of polymorphonuclear leukocytes; the erythrocytes were pelleted. Mononuclear and polymorphonuclear white cells from approximately 40% of the blood sample were processed for template DNA isolation (described above). The remainder was transferred to separate sterile tubes, contaminating erythrocytes were removed by treatment with an equal volume of red blood cell lysing buffer for 5 min at room temperature (hypotonic lysis), and the white cells were washed twice with an equal volume of phosphate-buffered saline. The leukocytes were pelleted each time by centrifugation at 5,000g for 10 min (4°C) in a fixed angle rotor, and were finally resuspended in a solution of 3 mM Tris-HCl buffer (pH 7.78 at room temperature), 2 mM EDTA, 2 mM dithiothreitol, 0.02 mM leupeptin, and 0.1 mM phenylmethylsulfonylfluoride.

Catalytic Properties and Thermostability of Human White Blood Cell and Recombinant NAT1 Variants. N-acetylation reactions were initially optimized for buffer type (Tris versus triethanolamine) and concentration, and the effect on catalytic activity of various agents such as KCl, dimethyl sulfoxide, bovine serum albumin, and butylated hydroxytoluene (Cribb et al., 1991; Grant et al., 1991; Dupret and Grant, 1992; Dupret et al., 1994; Hughes et al., 1998) was assessed. Since, in accordance with the ping-pong reaction mechanism of NAT, initial rates and steady-state kinetic constants will vary with the acetyl CoA concentration, catalytic experiments were done with the cofactor at 0.1 mM to enable comparison of the results with those in the literature (Grant et al., 1991; Dupret and Grant, 1992; Dupret et al., 1994; Hughes et al., 1998). Saturating amounts of the substrates were determined from kinetic experiments, and were 0.05 mM for 5-aminosalicylate; 0.1 mM for 4-aminosalicylate, 4-aminobenzoate, and 2-aminofluorene; 0.25 mM for p-phenetidine and p-anisidine; and 0.5 mM for sulfamethazine. 2-Aminofluorene was prepared in 1% methanol, with the solvent at a final concentration of 0.1% in the reaction mixtures (0.1% methanol was found to have no effect on 4-aminobenzoate and 4-aminosalicylate N-acetylation activities); all other substrates were aqueous solutions. The linearity of the reactions with cytosolic protein and time was established in comprehensive pilot experiments with each of the substrates and variants of heterologously expressed and human leukocyte NAT1.

Quantitation of N-Acetylated Products by HPLC. Parent compounds and arylamide metabolites were separated on a reversed phase C₁₈ column (Waters Nova-Pak C₁₈: 3.9 × 150 mm; 5-µm particle size), eluted at a flow rate of 1.0 ml/min under isocratic conditions. Elution was effected with a mobile phase of 0.1% aqueous trifluoroacetic acid/methanol (for 4-acetamidobenzoate, 4-acetamidosalicylate, and 5-acetamidosalicylate), 0.5% aqueous trifluoroacetic acid/methanol (for phenacetin and p-methoxyacetanilide), or 0.5% aqueous trifluoroacetic acid/methanol/acetonitrile (for 2-acetamidofluorene). The composition of the mobile phase was adjusted so that the retention time of the N-acetylated products was between 5.5 and 7 min. The UV/visible detector was set at the wavelength maximum of the arylamide metabolite, determined by spectral scanning with a Cary model 3E double beam spectrophotometer.

Analytical and Other Methods. Protein was determined by a dye-binding method with bovine serum albumin as the standard (Bradford, 1976). Statistical analysis was done by the unpaired Student's t test with the aid of the InStat program from GraphPad (San Diego, CA).

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Clarification of the Mutations in NAT1*11

Coding Region. A pronounced band compression resulting in the appearance of comigrating bands in adjacent lanes had been observed for palindromic nt 446-464 in NAT1*11 from subjects heterozygous for this allele (e.g., NAT1*4/*11), making it impossible to decide whether these were true heterozygosities or DNA secondary structure artifacts [see Table 1, footnote c, in Vatsis and Weber (1993)]. The compressed area in one of the original NAT1*11 samples (Vatsis and Weber, 1993) was fully disrupted upon sequencing in either direction with 7-deaza-dGTP or dITP, two dGTP analogs which form weaker base pairs that are readily denatured for polyacrylamide gel electrophoresis under regular conditions. Band compression was likewise eliminated when the same NAT1*11 sample was sequenced with dGTP, but submitted to electrophoresis on a polyacrylamide gel prepared with 40% formamide to effect stronger denaturing conditions (not shown). Since preparation of formamide gels is not always an easy task, the remainder of the original NAT1*11 samples and all new samples were routinely sequenced with dITP. Unlike the double allelic constitution of PCR-amplified genes that are sequenced directly (no subcloning) (Vatsis et al., 1991; Vatsis and Weber, 1993), only a single allele is introduced when a gene is cloned into a prokaryotic or eukaryotic host. It is noteworthy in this connection that no band compression was seen in the corresponding segment (nt 446-464) of NAT1*11 expressed in E. coli or COS-1 cells (de León, 1996), indicating that the structural abnormality at the palindrome was exacerbated by the presence of a second allele.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Structural variation in four human NAT1* alleles. The PCR-generated fragments of the variant NAT1* alleles (1.6 kb, nt 440 to 1175) were unambiguously identified by direct sequencing (no subcloning). The gene regions are marked as follows: stippled bar, 5' flank; hatched bar, 5' UTR; open bar, coding exon. Nucleotide and codon numbers are depicted below the bars. An AT-rich segment in the 3' UTR (nt 1065-1091) contains eight identical copies of AAT, ATA, or TAA motifs, as well as the consensus polyadenylation signal (AATAAA; nt 1086-1091, underlined). The highly conserved AATAAA hexanucleotide is obliterated in NAT1*10 by virtue of the T1088A transversion. The double underlining of this segment of NAT1*10 emphasizes that the AAT triplet (nt 1083-1085) immediately 5' to the mutated (AAAAAA) signal serves to generate a new and, according to the results of the present study, completely functional polyadenylation signal (AATAAA; nt 1083-1088). The end points of the nonanucleotide deleted from the 3' UTR of NAT1*11 have been identified by sequencing of the recombinant allele as nt 1077 and 1085. NAT1*16 has an insertion of three adenosines (AAA) immediately 5' to the AATAAA signal. Alleles NAT1*10, *11, and *16 share the C1095A transversion (position 1098 in NAT1*16).

3' UTR. Deletion of 9 bp from an AT-rich segment in the 3' UTR of PCR-generated NAT1*11 was also reported some time ago, but the actual location of the deletion site could not be ascertained because eight repetitive triplets (AAT, ATA, or TAA) within this segment gave 18 nonanucleotide combinations as potential deletion sites; elimination of any of these nonanucleotides resulted in the same sequence for NAT1*11 (Vatsis and Weber, 1993). Sequencing of the single allele of recombinant NAT1*11 expressed in E. coli or COS-1 cells (de León, 1996) has allowed a precise definition of the deletion end points as nt 1077 and 1088 (Fig. 1).

Determination of NAT1* Genotype

The NAT1* genotype of 26 unrelated individuals from five non-Caucasian groups was ascertained unambiguously by direct sequencing of both strands of 1.6-kb fragments (nt 440 to 1170) of PCR-generated NAT1*. The three structural variants (NAT1*4, *10, and *11) originally identified in 12 unrelated Caucasians (Vatsis and Weber, 1993) were also detected in the non-Caucasian population in this study together with a new allele (Vatsis et al., 1994), hereby designated NAT1*16 in compliance with the NAT nomenclature system (Vatsis et al., 1995). Mutations in NAT1*16 were confined to the 3' UTR, and included an insertion of three adenosines immediately 3' to the sixth position of the AATAAA signal (AATAAAAAA), as well as a C to A transversion four nucleotides downstream from the trinucleotide insertion (C¹⁰⁹⁵A, also displayed by NAT1*10 and *11) (Fig. 1). No other variant was found, and none could have gone undetected because allelic structure was determined by sequencing.

There are ten possible allelic combinations (genotypes) for four variant NAT1* alleles, and seven of these were detected despite the comparatively small size of the population (Table 1). By far the most prevalent genotypes were those of homozygotes and heterozygotes of NAT1*4, found in 33 of the 38 subjects. Genotypes with NAT1*10 were the second most abundant, accounting for 20 of the individuals (two homozygotes, 15 heterozygotes, and three compound heterozygotes), whereas NAT1*11 and *16 were present in only six subjects (three heterozygotes and three compound heterozygotes). Not surprisingly, the three genotypes not represented in the population were homozygotes and compound heterozygotes of the relatively infrequent NAT1*11 and *16 alleles.

NAT1*4 and *10 were widely distributed in all six ethnic groups, whereas NAT1*11 was present only in Caucasians and Asian Indians, and NAT1*16 was seen only in two Far Eastern groups (Japanese and mainland Chinese) (not shown). Given the limited number of subjects we have examined, it cannot be stated with certainty that NAT1*11 and *16 occur exclusively in Caucasians/Asian Indians and Orientals, respectively, nor can the allelic incidence in each of the six ethnic groups be accurately determined for the same reason. Preliminary estimates of the combined interethnic frequencies (n = 76 alleles) are possible, however, and show a distribution of 63% for NAT1*4, 29% for NAT1*10, 5% for NAT1*11, and 3% for NAT1*16 (see Table 1 for the allelic frequency computations). These frequency distribution estimates are quite close to those obtained in two other recent studies with subjects of mixed ethnic origin (Hughes et al., 1998; Lin et al., 1998).

Characterization of Recombinant Human NAT1* Variants

To enable an unobscured interpretation of the data, the effects of the coding region point mutations on NAT1 structure and function were evaluated separately from those of the mutations in the 3' UTR. This was accomplished by expression of the NAT1* coding region (nt 1-870) in E. coli, and of the coding region together with the 3' UTR (nt 1-1170) in COS-1 cells.

Expression of the Coding Region of NAT1*4 and *11 in E. coli. Since NAT1*4, *10, and *16 have identical coding sequences (Fig. 1), NAT1*4 was selected as representative of these alleles for transformation, along with the coding exon of NAT1*11, into E. coli. The expression level varied slightly (4-7%) for each of the expressed proteins in three separate bacterial inductions, and was approximately 25% lower for NAT1 11 [2.69 ± 0.09 units (NAT1 4) and 2.12 ± 0.05 units (NAT1 11) of immunoreactive NAT1/mg of E. coli cytosolic protein].

View this table: [in this window] [in a new window]   TABLE 2 Steady-state kinetics of N-acetylation by NAT1 4 and NAT1 11 isozymes expressed in E. coli strain BL21(DE3)pLysS Reaction mixtures with cytosol from transformed bacteria (1-2 µg of protein) were prepared in 20 mM Tris-HCl buffer (pH 7.5 at 37°C) containing an acetyl CoA regenerating system and the additional components described under Experimental Procedures. Mixtures with varying amounts of the substrate (2-100 µM) contained a fixed amount of acetyl CoA (0.1 mM); conversely, 4-aminosalicylate was held constant (0.1 mM) when the cofactor was varied from 0.005 to 0.5 mM. Kinetic constants with 4-aminosalicylate as the variable component were also determined with reaction mixtures prepared in 55 mM triethanolamine-HCl buffer (pH 7.5 at 37°C) containing 25 mM KCl and 2.5% (v/v) dimethyl sulfoxide; the acetyl CoA regenerating system and other components were as for the Tris-HCl buffer system. Reactions were at 37°C for 2 to 5 min, depending on the particular substrate or cofactor serving as the variable component, and arylamide products were quantitated by HPLC. NAT1 protein content was ascertained by densitometric scanning of 34-kDa bands on immunoblots, generated with a polyclonal human NAT1 antibody; 1 unit (U) of immunoreactive NAT1 was arbitrarily equated to 1000 relative band density units. Apparent Km and Vmax constants were calculated with the Cleland Hyper program (Cleland, 1967). The values (mean ± S.D.) represent data from three experiments (three separate bacterial inductions), with triplicate determinations in each experiment. Statistical analyses were performed with the unpaired Student's t test.

Expression of the Coding Region and 3' UTR of NAT1*4, *10, *11, and *16 in COS-1 Cells. The role of the 3' UTR in NAT1 expression and catalytic competence was investigated upon transient transfection of COS-1 cells with the coding exon and 290 bp of the 3' UTR of NAT1*4, *10, *11, and *16, respectively; all four alleles have dissimilar 3' UTR sequences (Fig. 1).

View this table: [in this window] [in a new window]   TABLE 4 Efficiency of transfection, NAT1 protein content, and specific N-acetylation activities in cytosol from COS-1 cells cotransfected with -galactosidase and NAT1* variants COS-1 cells (21 dishes per NAT1* allele) were transiently cotransfected with NAT1* recombinant construct (NAT1* coding and 3' noncoding exons) and the reporter vector pCR3Gal. The cells were pooled, disrupted by sonication, and the cytosolic fraction isolated by differential centrifugation of the homogenate. Catalytic activity of the cytosol was determined at 37°C with reaction mixtures in 20 mM Tris-HCl buffer, pH 7.5, containing 4-aminosalicylate (0.1 mM), 5-aminosalicylate (0.05 mM), or 2-aminofluorene (0.1 mM), acetyl CoA (0.1 mM), and the additional components detailed under Experimental Procedures. The activities are the mean ± S.D. of a total of 24 values obtained as follows: a) the average of the slopes of the lines constructed from data on protein and time dependence of the reactions, respectively [18 determinations: triplicates for each of three time points (5, 10, and 15 min with mixtures containing 1 µg of cytosol), and three amounts of cytosolic protein (0.5, 1, and 3 µg in mixtures incubated for 5 min)]; and b) the average of triplicates from each of two additional experiments conducted under the previously established initial rate conditions with respect to amount of cytosolic protein and reaction time (1 µg of protein irrespective of substrate; reactions carried out for 5 min with 2-aminofluorene and for 10 min with 4- and 5-aminosalicylate).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Quantitation of NAT1 protein content in COS-1 cell cytosol. Autoradiograms with varying amounts of COS-1 cell cytosolic protein (2.5-40 µg, each in triplicate) were generated with human anti-NAT1 as described under Experimental Procedures, scanned with an imaging densitomer, and the relative density of 34-kDa (NAT1) bands plotted against protein concentration. A direct proportionality is seen between NAT1 band density and protein up to 10 µg of cytosol containing NAT1 4, NAT1 10, or NAT1 11, and up to 20 µg of cytosol containing NAT1 16. Protein content is obtained from the slope of the linear segment of the curves.

Functional Correlates of NAT1 4 and NAT1 10 in Human White Blood Cells

The distribution of NATs in human leukocytes has been studied indirectly by steady-state kinetics with NAT1- and NAT2-specific substrates, and the findings suggested that NAT1 may be the sole acetylase in cytosol from mononuclear leukocytes and the principal acetylase in cytosol from neutrophils; the NAT1* genotype of the donors of the white blood cells was not evaluated in that study (Cribb et al., 1991).

Catalytic parameters and thermostability were investigated with the cytosolic fraction of mononuclear and polymorphonuclear leukocytes from two subjects homozygous for wild-type NAT1*4, and from one of the two NAT1*10 homozygotes detected in our population study (see Table 1). As demonstrated in Table 5, rates of N-acetylation of two NAT1-specific, two NAT1-selective, and one as yet uncategorized compound (p-anisidine; cf. Vatsis and Weber, 1997) were essentially identical for NAT1 4 and NAT1 10 in cytosol from either white blood cell type. By the same token, no differences were discernible in the apparent K_m [9-15 µM (NAT1 4); 9.5 µM (NAT1 10)] and V_max [7-10 nmol/min/mg of protein (NAT1 4); 8 nmol/min/mg of protein (NAT1 10)] determined with 4-aminosalicylate as the substrate. The K_i for the competitive inhibition exerted by amethopterin was the same for NAT1 4 (13 µM) and NAT1 10 (16.5 µM) in mononuclear leukocyte cytosol, and comparable to that obtained with the coding region of NAT1 4 (which is the same as that of NAT1 10 and NAT 1 16) expressed in E. coli cytosol (see Table 3).

Both NAT1 4 and NAT1 10 underwent rapid thermal inactivation, as evidenced by a 90-95% loss of 4-aminosalicylate N-acetylation activity after only 8 h of exposure of mononuclear leukocyte cytosol to 37°C. Unlike recombinant NAT1 in E. coli, thermal inactivation of naturally occurring NAT1 4 and NAT1 10 was a biphasic process. More importantly, the extent and rate constant associated with each phase of inactivation were the same for NAT1 4 and NAT1 10: about 35% of the activity catalyzed by either isozyme was lost in the first (more rapid) phase with a first order rate constant of 1 h¹, and the remaining two-thirds went in the second (slower) phase with a rate constant of 0.2 h¹. Inactivation of naturally occurring NAT1 4 and NAT1 10 was, therefore, two to ten times faster than that of the enzymes expressed in E. coli (~0.1 h¹; Table 3), quite possibly reflecting less than optimal folding or configuration of the NAT1 proteins in the artificial expression system.

Secondary Structure of the 3' UTR in pre-mRNAs for the Human NAT1 Variants

The secondary structure of the 3' UTR in the pre-mRNA for each of the NAT1 variants is characterized by a series of stems and loops (Fig. 3). Three structural elements are emphasized: the polyadenylation signal (AAUAAA; nt 1087-1096); the predicted cleavage site (GAA) 3' to the AAUAAA hexamer (between nt 1124 and 1125); and a U-rich sequence farther downstream (nt 1149-1152). The AAUAAA signal is present in bulges of the stem in NAT1 4, NAT1 10, and NAT1 16 pre-mRNAs, but in loop 3 of NAT1 11 pre-mRNA. The predicted cleavage site (GAA) is also located in a bulge in NAT1 4, NAT1 10, and NAT1 11, but found on the double-stranded stem of NAT1 16 pre-mRNA. The U-rich sequence is present on the single-stranded loop 4 in NAT1 4, NAT1 10, and NAT1 11, but is notably absent from NAT1 16 pre-mRNA. Both the predicted cleavage site and the U-rich downstream element appear to be sequestered in NAT1 16. Moreover, as a consequence of the altered secondary structure, the distance between the AAUAAA hexamer and the U-rich sequence is four nucleotides longer in NAT1 16 pre-mRNA (61 nt total) than in the pre-mRNA for the other three NAT1 variants (57 nt total).

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Predicted secondary structures of 3' UTR in the pre-mRNA for NAT1 4 (A), NAT1 10 (B), NAT1 11 (C), and NAT1 16 (D). The pre-mRNA secondary structures were predicted by the "Fold" program (Zuker and Stiegler, 1981) and "Squiggles" program (Osterburg and Sommer, 1981) provided by GenBank. "Fold" finds an optimal secondary structure of minimum free energy for an RNA molecule based on published values of stacking and loop destabilizing energies (Zuker and Stiegler, 1981). The three elements of interest, depicted in larger font and/or framed, include the polyadenylation signal (AAUAAA), the predicted cleavage site (GAA), and the U-rich sequence on loop 4.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Compensatory Elements in the 3' UTR of Human NAT1* Give Rise to a Fully Operational Polyadenylation Signal in NAT1*10

A T¹⁰⁸⁸A transversion in the third position of the polyadenylation signal (AATAAA; nt 1086-1091) of NAT1*10 obliterates this highly conserved element, which is situated at the tail end of an AT-rich segment of iterative AAT triplets (Fig. 1). Point mutations within the AATAAA hexamer are known to interfere markedly with the maturation and, consequently, promote the degradation of a host of functionally distinct mRNAs (Sheets et al., 1990), inferring that less NAT1 10 protein may be produced because of impaired 3' end processing of the corresponding transcript (Vatsis and Weber, 1993). We also considered that the AAT triplet (nt 1083-1085) immediately 5' to the first adenosine (nt 1086) of the mutated (AAAAAA) hexanucleotide could possibly serve to generate a new AATAAA signal (nt 1083-1088) in NAT1*10, but this would change the optimal distance between AATAAA and the cleavage/poly(A) addition site. Since there appears to be some spatial requirement for efficient mRNA production (Nevins, 1983), the functionality of a new polyadenylation signal with a 5' shift of three nucleotides could not have been accurately predicted for NAT1*10 (Vatsis and Weber, 1993), and, hence, had to be established experimentally.

The experiments in the present study clearly show that recruitment of the AAT triplet at positions 1083-1085 toward formation of a new AATAAA signal does not disrupt the predicted secondary structure of the pre-mRNA for NAT1 10, which is indistinguishable from that of wild-type NAT1 4 (Fig. 3, A and B). Consistent with this observation was the finding that the content of NAT1 10 protein in COS-1 cell cytosol was identical with that of wild-type NAT1 4. Furthermore, no differences existed in any of the biochemical parameters assessed with recombinant NAT1 4 and NAT1 10 in the eukaryotic host, and, more importantly, none was discernible with cytosol from mononuclear leukocytes of NAT1*4/*4 and NAT1*10/10 homozygotes. It may be concluded that the T¹⁰⁸⁸A transversion in NAT1*10 is well tolerated because the AAT triplet on the 5' side of the mutated polyadenylation signal does indeed generate a perfectly functional AATAAA hexamer. Our findings do not support the suggestion that NAT1 10 has greater N-acetylation activity than NAT1 4 (Bell et al., 1995), a conclusion also reached from NAT1 phenotypic assessment upon administration of 4-aminosalicylate to 148 unrelated subjects (Hughes et al., 1998).

The greater instability of NAT1 4 and NAT1 10 in their natural milieu (mononuclear leukocytes) as compared to the prokaryotic expression system appears to be the explanation for the substantially larger decrease in V_max seen with blood lysates from a heterozygote of NAT1*14 than with E. coli cytosol harboring the coding region of NAT1*14 (Hughes et al., 1998). Negative modulation of the catalytic activity by the 3' UTR mutations in NAT1*14 cannot explain the discrepant V_max results of Hughes et al. (1998), since the data presented above (Tables 4 and 5) amply demonstrate that NAT1*10, which has the same polyadenylation signal mutation as NAT1*14, encodes a protein with catalytic and other properties indistinguishable from those of wild-type NAT1 4.

Coding and 3' UTR Mutations in NAT1*11 Do Not Significantly Affect Function

Maximal rates of substrate N-acetylation were comparable for recombinant wild-type NAT1 4, with a coding sequence superimposable to that of NAT1 10 and NAT1 16, and for mutant NAT1 11, the only variant with missense mutations among the four identified in this study. NAT1 4 and NAT1 11 also exhibited the same sensitivity to the competitive inhibitor amethopterin, and possessed similar energies of activation and thermal stability assessed with 4-aminosalicylate as the substrate. On the other hand, NAT1 11 was invariably found to have a greater catalytic efficiency by virtue of a 1.5- to 2-fold higher affinity for all arylamines examined, a statistically significant effect observed irrespective of reaction buffer or expression host.

One of the mutational events in NAT1*11 concerns a sizeable 9-bp deletion (nt 1077-1088) from the same AT-rich segment containing the polyadenylation signal (nt 1064-1091; Fig. 1). AT-rich stretches have been found in the 3' UTR of many transcripts with high turnover rates, and are referred to as destabilizing motifs because they facilitate mRNA degradation and also act as negative modifiers of mRNA translation. The corollary is that elimination of nine bases from a destabilizing sequence should result in higher steady-state levels of NAT1 11 transcript and protein (Vatsis and Weber, 1993). More recent studies have shown, however, that only a specific AT-rich sequence, namely, TTATTTAT, can act as a destabilizing unit to bring about rapid mRNA decay (Zubiaga et al., 1995). Human NAT1* alleles do not contain the specific TTATTTAT element, and the identical immunoreactivity of NAT1 4, NAT1 10, and NAT1 11 in COS-1 cell cytosol is consistent with the absence of such a motif, as is the virtually identical secondary structure of the pre-mRNAs for these isozymes, as well as the similarity in catalytic activities and V_max constants of NAT1 4 and NAT1 11.

Collectively, the prokaryotic and eukaryotic expression studies have shown no major differences in catalytic or other properties of wild-type NAT1 4 and mutant NAT1 11, and others have obtained overlapping values for NAT1*4/*11 heterozygotes and NAT1*4/*4 homozygotes phenotypically evaluated with 4-aminosalicylate in vivo (Hughes et al., 1998). It should be pointed out, however, that the observed similarity in N-acetylation activities of NAT1 4 and NAT1 11 with the substrates examined in this study, including three simple amine drug molecules and one carbocyclic arylamine carcinogen (2-aminofluorene), does not in any way preclude differences in catalytic activity with other environmental/occupational carbocyclic carcinogens and dietary heterocyclic mutagens, which are of particular importance to chemical carcinogenesis and cancer epidemiology (Lin et al., 1998).

Formation of NAT1 16 Protein is Obstructed by Structural Modification in the 3' UTR

Numerous publications have documented the pivotal role of RNA secondary structure in the assembly of a stable formation complex, a stepwise process involving the cooperative binding of six nuclear factors to a pre-mRNA molecule, which is absolutely essential for efficient 3' end processing of the pre-mRNA. A mutationally modified pre-mRNA may promote destabilization of the formation complex, thereby adversely affecting maturation. This topic is reviewed extensively by de León (1996), and a synopsis is presented in the ensuing paragraph.

Consensus sequences in target elements of RNA-binding proteins are frequently found in loops, bulges, or interior loops of the RNA, representing single-stranded regions for component binding (McCarthy and Kollmus, 1995). Additionally, the AATAAA signal in many viruses and certain mammalian mRNAs is flanked by nucleotide segments that form base-pairs, giving rise to a hairpin structure with the AATAAA signal located in the single-stranded loop. Cleavage efficiency was diminished 50% by disruption of the stem structure, and 90% by a decrease in the size of the loop (Berkhout et al., 1995). Cleavage and polyadenylation efficiencies were likewise severely curtailed when a sequence insertion between the AATAAA and GT/T elements, as seen in NAT1*16, failed to form a stem-loop structurally similar to these two 3' UTR elements. Poly(A) sites can, therefore, tolerate sequence insertions only if the inserted sequence forms RNA secondary structures that resemble the AATAAA and GT/T elements (Brown et al., 1991). Nuclease accessibility/sensitivity to the cleavage site is also affected by alterations in stem-loop structures brought about by mutation (cf. de León, 1996).

The insertion of three adenosines on the 3' side of the AATAAA signal in NAT1*16 resulted in parallel (2-fold) decreases in COS-1 cell cytosolic content of NAT1 16 protein and specific N-acetylation activity with three substrates, indicating that the structural alteration in the 3' UTR was deleterious solely to the synthesis, not the catalytic competence, of NAT1 16 protein. The structural alterations resulting from the AAA insertion in the 3' UTR of NAT1*16 (Fig. 1) entail obliteration or drastic reduction in the size of loop 4, sequestration of the predicted cleavage site, and a longer distance between the polyadenylation signal and the downstream T-rich element (compare Fig. 3A and 3D). It follows from the discussion elaborated in the preceding paragraph that these structural modifications may have several untoward consequences, such as decreased binding affinity of cleavage stimulation factor for the T-rich element because of the apparent sequestration or marked reduction of this element; less accessibility/sensitivity of the predicted cleavage site to endonucleases; diminished cooperativity among the various nuclear factors, resulting from a lower binding affinity of cleavage stimulation factor for the T-rich motif, as well as from the increased distance between AATAAA and the T-rich motif; and/or assembly of an unstable formation complex that is detrimental to transcript maturation. Suffice it to say that, whatever the precise mechanism, structural abnormalities of this sort cause disruption of the secondary structure of the pre-mRNA for NAT1 16, culminating in the observed decrease in the amount and corresponding decrease in the catalytic activity of NAT1 16 protein in COS-1 cell cytosol. It should also be mentioned at this juncture that, since mRNA levels were not measured, other possibilities for the decreased immunoreactivity of NAT1 16, such as impaired translation efficiency, cannot be ruled out at this time.