INTRODUCTION

Features of proteins that confer metabolic instability are called degradation signals, or degrons (1). The essential component of one degradation signal, termed the N-degron, is a destabilizing N-terminal residue of a protein (2). A set of N-degrons containing different destabilizing residues yields a rule, termed the N-end rule, that relates the in vivo half-life¹ of a protein to the identity of its N-terminal residue (reviewed in Refs. 3 and 4). The N-end rule pathway has been found in all species examined, including the eubacterium Escherichia coli (5), the yeast Saccharomyces cerevisiae (6), and mammalian cells (7, 8). The N-end rules of these organisms are similar but distinct (4).

The understanding of the functions of the N-end rule is sketchy. In particular, the N-end rule pathway has been shown to be required for the import of peptides in S. cerevisiae (9); it may also be involved in the control of signal transduction and cell differentiation (10, 11). Physiological substrates of the N-end rule pathway include Sindbis virus RNA polymerase (12) and yeast G, the GPA1-encoded subunit of the heterotrimeric G protein (13).

In eukaryotes, the N-degron comprises at least two determinants: a destabilizing N-terminal residue and an internal lysine (or lysines) of a substrate (6, 14, 15, 16). The Lys residue is the site of formation of a multiubiquitin chain (17, 18, 19, 20). Ubiquitin (Ub)² is a 76-residue protein whose covalent conjugation to other proteins is involved, directly or by way of regulation, in a multitude of processes, including cell growth and differentiation, signal transduction, DNA repair, and the transport of substances across membranes. In many of these processes, Ub acts through routes that involve protein degradation (reviewed in Refs. 4 and 21, 22, 23, 24, 25, 26).

The N-end rule is organized hierarchically. In eukaryotes such as the yeast S. cerevisiae, Asn and Gln are tertiary destabilizing N-terminal residues (denoted as N-d^t) in that they function through their conversion, by N-terminal amidohydrolase (Nt-amidase), into the secondary destabilizing N-terminal residues Asp and Glu (denoted as N-d^s), whose destabilizing activity requires their conjugation, by Arg-tRNA-protein transferase (R-transferase), to Arg, one of the primary destabilizing N-terminal residues (denoted as N-d^p) (6, 27, 28). The N-d^p residues are bound directly by N-recognin (also called E3), the recognition component of the N-end rule pathway (4). In S. cerevisiae, N-recognin is a 225-kDa protein (encoded by UBR1) that selects potential N-end rule substrates by binding to their N-d^p residue Phe, Leu, Trp, Tyr, Ile, Arg, Lys, or His (29, 30, 31). This binding is followed by the formation of a substrate-linked multi-Ub chain in a reaction mediated by the Ub-conjugating enzyme (E2) Ubc2p, one of at least 11 distinct E2 enzymes in S. cerevisiae (31, 32). The substrate is then processively degraded by the 26 S proteasome, an ~2-MDa, ATP-dependent, multisubunit protease (33, 34, 35, 36, 37, 38). The four ``upstream'' components of the S. cerevisiae N-end rule pathway (Nta1p (Nt-amidase), Ate1p (R-transferase), Ubc2p (a Ub-conjugating enzyme), and Ubr1p (N-recognin)) are physically associated in a targeting complex (27, 31).³

The NTA1-encoded Nt-amidase of S. cerevisiae can deamidate N-terminal Asn or Gln (3, 27). Stewart et al. (39, 40) reported the purification of a porcine Nt-amidase and the cloning of its ~1.3-kb cDNA. In contrast to the yeast Nt-amidase (Nta1p), this Nt-amidase, termed Nt^N-amidase, can deamidate N-terminal Asn but not Gln. Described below is the isolation and characterization of Ntan1, an ~17-kb gene⁴ that encodes a mouse homolog of the porcine enzyme, termed Nt^N-amidase (the superscript ``N'' and the second ``N'' in the gene's name refer to the Asn specificity of this Nt-amidase).

Both Asn and Gln are destabilizing residues in the mammalian N-end rule (7, 8). Furthermore, both N-terminal Asn and Gln of test proteins are deamidated in mammalian cell extracts (7).⁵ Since Nt^N-amidase deamidates N-terminal Asn but not Gln, there must exist yet another mammalian Nt-amidase, which can deamidate N-terminal Gln. The resulting bifurcation in the structure of the mammalian N-end rule is illustrated in Fig. 1. Ntan1 is the first cloned mammalian gene whose function appears to be confined to the N-end rule pathway.

EXPERIMENTAL PROCEDURES

The E. coli strains MC1061, DH5, and JM101 were grown in Luria broth containing relevant antibiotics and were transformed using the calcium chloride method (41). S. cerevisiae was grown on rich (YPD) or synthetic medium containing 2% glucose (SD medium) or 2% galactose (SG medium) (42). Unless stated otherwise, the transformation of S. cerevisiae was carried out as described (43). The S. cerevisiae strains were YPH500 (MAT ura3-52 lys2-801 ade2-101 trp1-63 his3-200 leu2-1) (44) and SGY4, an nta1-2::LEU2 derivative of YPH500. SGY4 was constructed by transforming YPH500 (using the procedure of Ref. 45) with an XhoI-XbaI fragment of the plasmid pSG8 that bore the nta1-2::LEU2 deletion/disruption allele and selecting for Leu⁺ cells. Southern hybridization analysis of Leu⁺ transformants, using restriction endonuclease cuts diagnostic of the transplacement (46), was used to verify the predicted structure of the integrated nta1-2::LEU2 allele. The plasmid pSG8 was derived from a plasmid (based on pRS316 (41)) that contained S. cerevisiae NTA1 between XhoI and XbaI sites (27). Replacement of the SpeI-EcoRI fragment of the NTA1 ORF with LEU2 yielded pSG8.

-Galactosidase Assays

Colony assays for E. coli -galactosidase (gal) in S. cerevisiae were carried out by overlaying yeast colonies on SG plates with 0.5% agarose containing 0.1% SDS, 4% dimethylformamide, and a 0.1 mg/ml concentration of the chromogenic gal substrate 5-bromo-4-chloro-3-indolyl -D-galactoside, followed by incubation for a few hours at 37 °C and examination of the colony color (white to blue, depending on the level of gal). Quantitative assays for gal activity were carried out with S. cerevisiae whole cell extracts using another gal substrate, chlorophenol red -D-galactopyranoside. Cells in a 1-ml culture (A₆₀₀ ~ 1) were pelleted by centrifugation and resuspended in 1 ml of buffer K (1 mM MgCl₂, 0.1 M potassium phosphate, pH 7.0). After determining the A₆₀₀ of the suspension, 50- or 100-µl samples were diluted to 1 ml with buffer K. 0.1% SDS (20 µl) and CHCl₃ (50 µl) were then added; the suspension was vortexed for 20 s and incubated for 15 min at 30 °C, followed by the addition of 20 µl of chlorophenol red -D-galactopyranoside (34 mg/ml in buffer K) and further incubation at 30 °C. The A₅₇₄ of a sample was measured at different incubation times (after clarification by a brief centrifugation in a microcentrifuge). The chlorophenol red -D-galactopyranoside units (U_CPRG) of gal activity were calculated as follows: U_CPRG = A₅₇₄/t·v·A₆₀₀, where t and v are the time of incubation (min) and the sample volume (ml), respectively.

The mouse Friend erythroleukemia cell line MEL-C19 was provided by Dr. L. A. Brents (NCI, National Institutes of Health, Bethesda, MD). The mouse lymphoma (EL4.IL-2) and myoblast (C₂C₁₂) cell lines were purchased from the American Type Culture Collection (Rockville, MD). MEL and EL4 cells were grown in Dulbecco's modified Eagle's/Ham's F-12 medium (Mediatech, Herndon, VA). C₂C₁₂ cells were grown in Dulbecco's modified Eagle's medium plus 4.5 g/liter glucose (Mediatech). All media were supplemented with 10% fetal bovine serum, glutamine, penicillin, streptomycin, and pyruvate (47). MEL cells were induced to differentiate in RPMI 1640 medium (Mediatech) containing 10% fetal bovine serum by adding N,N-hexamethylenebisacetamide to a final concentration of 3 mM and incubating a culture for another 5 days. C₂C₁₂ cells were induced to differentiate by replacing fetal bovine serum with 10% horse serum and incubating a confluent culture for 4-5 days.

Stewart et al. (39, 40) have purified an amidohydrolase, termed Nt^N-amidase, from porcine liver that is specific for N-terminal Asn in polypeptides; they also determined the N-terminal sequence of porcine Nt^N-amidase (peptide 1) and several sequences of internal peptides (including peptide 2) produced by fragmentation with cyanogen bromide (see Fig. 2A). Stewart et al. (40) employed ``reverse-translated'' oligonucleotides corresponding to the sequences of peptides 1 and 2 to isolate a 632-bp porcine cDNA fragment whose 5- and 3-regions encoded the sequences of peptides 1 and 2, respectively.

To clone a mouse cDNA (termed Ntan1 cDNA) that encoded a homolog of the porcine Nt^N-amidase, we used cDNA libraries (CLONTECH, Palo Alto, CA) prepared from mouse liver (in gt11) and MEL-C19 cells (in gt10). A low-stringency hybridization screening (carried out as recommended by CLONTECH) of the mouse liver cDNA library utilized the above 632-bp porcine cDNA fragment (at the time, only that fragment of the subsequently cloned porcine Ntan1 cDNA had been isolated by Stewart et al. (40)). A positive clone contained a 1560-bp mouse DNA insert; a 114-bp region of this insert was 85% identical to the nucleotide sequence of a region of the porcine Ntan1 mRNA. This and other evidence (see ``Results'') suggested that the 1560-bp mouse DNA fragment was derived from an unspliced or partially spliced Ntan1 pre-mRNA.

Since a Northern hybridization with the 1560-bp fragment as a probe suggested that the abundance of mouse Ntan1 mRNA in MEL cells was significantly higher than in the liver, the 1560-bp fragment was used to screen a gt10-based cDNA library prepared from MEL cells. Two positive clones (out of ~5 × 10⁶ clones screened) containing mouse cDNA inserts of 1030 and 605 bp were obtained. The amino acid sequence encoded by the larger insert (which encompassed most of the 605-bp insert) was 83% identical to the deduced sequence of porcine Nt^N-amidase (40); the region of similarity began 33 nucleotides downstream of the ATG start codon in the porcine Ntan1 ORF. In the 605-bp mouse cDNA insert, the TAA stop codon of the putative mouse Ntan1 ORF was followed by 157 nucleotides of untranslated region linked to a poly(A) tract. The 1030-bp mouse cDNA insert ended 9 bp upstream of the poly(A) addition site.

On the assumption that the 1030-bp mouse cDNA insert lacked a sequence corresponding to the 5-coding region of the mouse Ntan1 mRNA, we employed the 5-RACE-PCR technique (48, 49) to amplify the missing region, using an oligonucleotide primer specific for the 1030-bp clone and a primer complementary to an in vitro produced 5-oligo(dC) tract. Specifically, poly(A)⁺ mRNA was isolated from ~4 × 10⁸ MEL cells using the Fast Track mRNA isolation kit (Invitrogen, San Diego, CA) and dissolved in 10 mM Tris-HCl, 1 mM EDTA, pH 8.0, to an A₂₆₀ of 5. 1.5 µl of this sample were diluted with 12 µl of diethyl pyrocarbonate-treated water, incubated at 70 °C for 5 min, and cooled on ice. To this sample were added 20 pmol of a primer corresponding to the antisense strand of the mouse Ntan1 ORF between nucleotides +700 and +725 (GenBank^TM/EMBL accession number U57692[GenBank]), 2 µl of 10 × PCR buffer (Perkin-Elmer), 2 µl of 0.1 M dithiothreitol, 1 µl of 10 mM dNTPs, 1 µl of 25 mM MgCl₂, and 1 µl of bovine serum albumin (2 mg/ml). The sample was incubated at 42 °C for 2 min, followed by the addition of 1 µl of Superscript II reverse transcriptase (Life Technologies, Inc.) and incubation at 42 °C for another 40 min. The temperature was increased to 55 °C, followed by the addition of 1 µl of RNase H (Life Technologies, Inc.; 2 units/ml) and incubation for 20 min. The resulting cDNA products were purified with Glass Max (Life Technologies, Inc.) using three standard washes plus a wash with cold 70% ethanol and were eluted with 50 µl of water.

To produce a cDNA-linked 5-oligo(dC) extension, 5 µl of purified cDNA were diluted with 11.5 µl of water; incubated at 70 °C for 5 min; cooled on ice; and then mixed with 1 µl of 10 × PCR buffer (to a final concentration of 0.5×), 1 µl of 25 mM MgCl₂, 0.5 µl of bovine serum albumin (2 mg/ml), 0.5 µl of 10 mM dCTP, and 0.25 µl of terminal transferase (Boehringer Mannheim). After incubation at 37 °C for 5 min, the enzyme was inactivated by heating the sample at 65 °C for 10 min. The first round of RACE-PCR amplification was carried out in a 100-µl sample containing 10 µl of 10 × PCR buffer, 2.5 mM MgCl₂, 5 µl of cDNA linked to oligo(dC), 20 pmol of a primer corresponding to the antisense strand of the Ntan1 ORF between nucleotides +283 and +310 (GenBank^TM/EMBL accession number U57692[GenBank]), and 20 pmol of G-anchor primer (5-AGGCCACGCGTCGACTAGTAC(G)₁₇-3). The sample was incubated at 94 °C for 5 min and then at 57 °C for 8 min, followed by the addition of AmpliTaq DNA polymerase (Perkin-Elmer) and incubation at 72 °C for 8 min to produce the strand complementary to cDNA. Thereafter, 35 cycles of a three-step PCR amplification were carried out. Each step involved consecutive incubations for 30 s at 94 °C, for 1 min at 57 °C, and for 2 min at 72 °C. 2 µl of the first-round PCR product were used for the second round of PCR that utilized, in the same total volume, 0.4 nmol of G-adapter primer (same as the G-anchor primer but lacking G₁₇) and 0.4 nmol of a primer corresponding to the antisense strand of the Ntan1 ORF between nucleotides +232 and +259.

The PCR-produced DNA fragments were inserted into pCRII (Invitrogen). Four of the resulting clones were sequenced; all of them contained an ATG codon preceded by one and the same 34-bp sequence and followed by a region encoding an amino acid sequence similar to the directly determined N-terminal sequence of the purified porcine Nt^N-amidase (40). An in-frame stop codon 6 bp upstream of the above ATG codon suggested that the latter was indeed the in vivo start codon of the Ntan1 ORF. To construct the full-length Ntan1 cDNA, PCR was carried out using the above 1030-bp Ntan1 cDNA fragment as a template, the primer 5-GTAGGTACCGCCTTTGCCAAAAATAAGATTTTATTTTG-3 (which was complementary to the 3-end of Ntan1 cDNA and contained a KpnI site), and the primer 5-ATCCTCGAGCATATGCCACTGCTGGTGGATGGGCAGCGCGTCCGTCTGCCACGGTCCGC-3 (which was complementary to the 5-end of the 1030-bp fragment and contained an XhoI site as well as the RACE-PCR-derived 5-proximal region of Ntan1 cDNA that was absent from the 1030-bp fragment). The PCR product was digested with KpnI and XhoI; the resulting fragment was inserted into KpnI/XhoI-cut TRP vector (a gift from Dr. S. Elledge, Baylor College of Medicine, Houston, TX), a variant of the vector YES (50) that contained S. cerevisiae TRP1 (instead of URA3) as a selectable marker. The resulting plasmid, pSG61, was used to express the mouse NTAN1 protein in S. cerevisiae from the P_GAL1 promoter.

A genomic DNA library produced from mouse embryonic stem cells (strain C129) and carried in the P1 phage (51) was used. Screening of the library (164,000 P1 clones arranged in 400 pools) was carried out by Genome Systems Inc. (St. Louis, MO) as a custom service, using PCR and synthetic primers derived from the sequence of the mouse Ntan1 cDNA (nucleotides +641 to +663 and nucleotides +730 to +752 (GenBank^TM/EMBL accession number U57692[GenBank])) (see Fig. 2B). Four positive clones were identified; one of them (clone 1798) was used for further analysis. To amplify and isolate a P1 plasmid containing the insert 1798, 0.3 ml of an overnight E. coli NS3529 culture bearing the plasmid was inoculated into 30 ml of Luria broth containing kanamycin at 25 µg/ml. Following a 30-min incubation at 37 °C in a rotary shaker, isopropylthiogalactoside was added to a final concentration of 1 mM (to induce the lytic operon of P1), and the culture was incubated at 37 °C until saturation (~5 h). The P1 plasmid was isolated using the alkaline lysis method (41). The plasmid was digested with various restriction endonucleases; the resulting fragments were subcloned into pUC18 (41) or Bluescript II SK(+) (Stratagene, La Jolla, CA) and propagated in E. coli DH5. Genomic DNA inserts containing sequences of Ntan1 cDNA were identified by colony hybridization using the Ntan1 cDNA insert of pSG61 (see above) as a probe. The regions of Ntan1 that encompassed exons I-X and the regions of introns adjacent to exons were sequenced using the chain termination method (41) and exon-specific primers. The physical map of the Ntan1 locus (see Fig. 3) was produced initially by restriction mapping of Ntan1-specific genomic DNA fragments and by PCR-based analyses of DNA that spanned the borders of sequenced regions. Subsequently, the entire ~17-kb Ntan1 gene was sequenced using a similar strategy: identifying a set of restriction fragments of P1 clone 1798 that encompassed the Ntan1 locus, subcloning these fragments to a size of ~1.5 kb into Bluescript II SK(+), and sequencing the inserts from both ends. The remaining regions were sequenced using primers specific for the junctions to already sequenced fragments. Most (>75%) of the Ntan1 fragments were sequenced on both strands. These nucleotide sequences have been submitted to the GenBank^TM/EMBL Data Bank with accession numbers U57692[GenBank] (Ntan1 cDNA) and U57691[GenBank] (Ntan1 genomic DNA).

We screened the gt10-based cDNA library prepared from MEL cells (see above) with the rabbit E2_14K cDNA (a gift from Dr. S. Wing) (52) as a probe. In this library, the E2_14K probe detected an ~30-fold higher number of clones than the Ntan1 probe, consistent with a lower in vivo level of Ntan1 mRNA in comparison with E2_14K mRNA (see ``Results''). All four of the analyzed mouse E2<sub>14K</sub> cDNAs contained identical sequences encompassing the E2<sub>14K</sub> ORF (GenBank<sup>TM</sup> accession number U57690[GenBank]), which was 93% identical to the rabbit E2<sub>14K</sub> ORF and which encoded an amino acid sequence 100% identical to that of rabbit E2<sub>14K</sub>. However, while three of the mouse E2<sub>14K</sub> cDNAs contained a poly(A) sequence 174 bp downstream of the E2<sub>14K</sub> ORF, in the fourth cDNA, the poly(A) sequence was located 500 bp farther downstream (data not shown), consistent with the presence of two E2<sub>14K</sub> mRNA species in the Northern hybridization patterns (see ``Results'').

5-Mapping of Ntan1 mRNA

Total cytoplasmic RNA was isolated from mouse MEL-C19 cells as described (41). The primer extension mapping was carried out as described (41) using 50 µg of RNA/assay. The Superscript II polymerase (Life Technologies, Inc.) was used for the reverse transcription step, which employed the following synthetic primers: nucleotides 985-1014 and 1485-1511, corresponding to the antisense DNA strand of Ntan1 (see Fig. 2B). Reaction products were analyzed by polyacrylamide gel electrophoresis on a 9% sequencing gel.

The chromosomal position of Ntan1 was determined using the interspecific backcross analysis. Interspecific backcross progeny were generated by mating (C57BL/6J × Mus spretus)F1 females and C57BL/6J males as described (53). A total of 205 N₂ mice were used to map Ntan1. DNA isolation, digestion with restriction endonucleases, gel electrophoresis, and Southern hybridization were performed essentially as described (54). All blots were prepared with Hybond-N⁺ nylon membranes (Amersham Corp.). The probe, an ~2.3-kb EcoRI fragment of mouse genomic DNA that encompasses exons II-IV of Ntan1, was labeled with [-³²P]dCTP using a nick translation labeling kit (Boehringer Mannheim); washing was carried out to a final stringency of 0.1 × SSCP, 0.1% SDS at 65 °C. An ~15-kb fragment and an ~4-kb fragment were detected with this probe in HindIII-digested C57BL/6J DNA and M. spretus DNA, respectively. The presence or absence of the ~4-kb M. spretus-specific HindIII fragment was followed in DNA of backcross mice.

A description of the probes and restriction fragment length polymorphisms for the loci linked to Ntan1, including the loci encoding protamine-1 (Prm1), CCAAT/enhancer-binding protein- (Cebpd), and the immunoglobulin chain (Igl), has been reported (55). Recombination distances were calculated as described (56) using the program SPRETUS MADNESS. Gene order was determined by minimizing the number of recombination events required to explain the allele distribution patterns.

For Southern hybridization, DNA was isolated from mouse liver, digested with restriction endonucleases, fractionated on 1% agarose gels, and blotted onto Hybond-N⁺ membranes as described (41). Prehybridization was carried out at 42 °C for 5 h in a buffer containing 2 × Denhardt's solution, 1 M NaCl, 0.5% SDS, 1 mM EDTA, 50 mM Tris-HCl, pH 7.5, and 0.1 mg/ml sheared denatured calf thymus DNA. Hybridization was performed at 37 °C for 40 h in hybridization buffer (0.5 × Denhardt's solution, 30% formamide, 1% SDS, 5 mM EDTA, 1 M NaCl, 50 mM sodium phosphate, pH 7.2, and 0.1 mg/ml sheared denatured calf thymus DNA). The probe DNA was labeled to ~10⁹ cpm/µg using [³²P]dCTP and the Random Primer labeling kit (DuPont NEN). The final concentration of labeled, heat-denatured probe DNA was ~5 ng/ml. After hybridization, the membranes were washed under low-stringency conditions (2 h at 25 °C in 0.4 M NaCl, 1% SDS, 5 mM EDTA, and 5 mM sodium phosphate, pH 7.2). For washes of increasing stringency, carried out with shaking for 30 min, the concentration of NaCl was decreased in 0.1 M increments; at 0 M NaCl, the temperature of washes was increased, in 10 °C increments, to 65 °C.

For Northern hybridization, the poly(A)⁺ RNA was isolated from mouse MEL-C19 cells, mouse lymphoma EL4.IL-2 cells, and mouse C₂C₁₂ myoblasts using the Fast Track mRNA isolation kit (Version 3.5). The mouse skeletal muscle poly(A)⁺ RNA was purchased from CLONTECH. RNA was fractionated by electrophoresis on 1% agarose-formaldehyde gels and blotted onto Hybond-N⁺ membranes as described (41). Northern hybridization was carried out as recommended by CLONTECH (protocol c, for hybridization with DNA probes). The probes (labeled with ³²P as described above) were mouse -actin cDNA (CLONTECH) and DNA inserts of the plasmids pSG61 (mouse Ntan1 cDNA; see above), pSG72 (mouse E2_14K cDNA; see above), and pSG76 (mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA). The GAPDH cDNA was produced using reverse transcription-PCR (41), the oligo(dT)-primed MEL-C19 cDNA library (see above), and the GAPDH-specific primers 5-ATGGTGAAGGTCGGTGTGAACGGA-3 and 5-TTACTCCTTGGAGGCCATGTAGGC-3, derived from the sequence of GAPDH cDNA (57).

ORFs encoding Ub-X-dihydrofolate reductase fusion proteins (Ub-X-DHFR proteins) (construct I of Ref. 6; X = Asn, Asp, or Gln) were used to replace ORFs encoding Ub-X-gal in pKKUB-X-gal plasmids (5). The resulting plasmids, pSG4, pSG41, and pSG44, expressed Ub-Asn-DHFR, Ub-Asp-DHFR, and Ub-Gln-DHFR, respectively, from the P_trc promoter. E. coli JM101 was transformed with one of these plasmids and also with pJT184, which expressed Ubp1p, a Ub-specific protease of S. cerevisiae (5). A 50-ml culture was grown at 37 °C to an A₆₀₀ of ~0.9 in Luria broth plus ampicillin (40 µg/ml) and chloramphenicol (20 µg/ml). Unless stated otherwise, all subsequent procedures were carried at 4 °C. The culture was chilled on ice for 15 min; centrifuged at 3000 × g for 5 min; washed twice with M9 medium (41); and resuspended in 50 ml of M9 medium supplemented with glucose (0.2%), thiamine (2 µg/ml), ampicillin (40 µg/ml), 1 mM isopropylthiogalactoside, and methionine assay medium (Difco). The suspension was shaken for 1 h at 37 °C, followed by the addition of 1 mCi of Tran³⁵S-label (ICN, Costa Mesa, CA) and further incubation for 1 h at 37 °C. Unlabeled L-methionine was then added to 1 mM, and the shaking was continued for another 10 min. The cells were harvested by centrifugation; washed twice with M9 medium; and resuspended in 0.5 ml of 25% (w/v) sucrose, 50 mM Tris-HCl, pH 8.0. Freshly prepared egg white lysozyme (0.1 ml of a 10 mg/ml solution in 0.25 M Tris-HCl, pH 8.0) was then added, and the suspension was incubated for 5 min at 0 °C, followed by the addition of 0.1 ml of 0.5 M EDTA, pH 8.0, and a 5-min incubation at 0 °C. The suspension was transferred to a centrifuge tube containing 1 ml of 65 mM EDTA, 50 mM Tris-HCl, pH 8.0, and protease inhibitors (25 µg/ml each antipain, chymostatin, leupeptin, pepstatin, and aprotinin; Sigma). Triton X-100 (10 µl of a 10% (v/v) solution) was then added and dispersed by pipetting. The lysate was centrifuged at 40,000 × g for 30 min in a TL-100 ultracentrifuge (Beckman Instruments). The supernatant containing ³⁵S-X-DHFR (the Ub moiety of Ub-X-DHFR was removed in vivo by the Ubp1p protease) was frozen in liquid N₂ and stored at 80 °C. X-DHFR (X = Asn, Asp, or Glu) was purified on a methotrexate affinity column (Pierce; 0.5-ml bed volume) that had been equilibrated in 50 ml of 50 mM Tris-HCl, pH 7.2. The thawed ³⁵S-supernatant was clarified by centrifugation at 12,000 × g for 1 min and applied to the column, which was then washed with 10 ml of 1 M KCl, 20 mM Tris-HCl, pH 8.0. X-DHFR was eluted with 2 mM folic acid, 50 mM Tris-HCl, pH 8.0. The elution buffer was applied in 1-ml samples, and 1-ml fractions were collected. Pooled fractions containing the peak of eluted ³⁵S were dialyzed for 20 h against two changes of 50% glycerol, 1 mM MgCl₂, 0.1 mM EDTA, and 40 mM HEPES, pH 7.5, and stored at 20 °C. ³⁵S-X-DHFR proteins were examined by SDS-polyacrylamide gel electrophoresis and fluorography and found to be >95% pure.

30 ml of an S. cerevisiae culture (A₆₀₀ ~ 1) were collected; washed once with 40 ml of water; pelleted at 3000 × g; washed once with 40 ml of buffer A (125 mM KCl, 15 mM MgCl₂, and 70 mM Tris-HCl, pH 8.0); and resuspended in 1 ml of buffer A containing 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, and the above mixture of protease inhibitors. 0.4 ml of 0.5-mm glass beads was added, and the suspension was vortexed at the maximal setting (three times for 1 min, with 2-min incubations on ice in between). The lysate was centrifuged at 12,000 × g for 5 min, and the supernatant was used for the amidase assay either immediately or after dilution with buffer A containing 1% bovine serum albumin.

In most assays, 5 µl of ³⁵S-X-DHFR (0.5 mg/ml in storage buffer) were mixed with 20 µl of yeast extract, incubated for 20 min at 30 °C, and thereafter chilled on ice. Samples (5 µl) were applied onto isoelectric focusing (IEF) polyacrylamide plates, pH 3.5-9.5 (Pharmacia Biotech Inc.), precooled to 10 °C. IEF was carried out for 80 min at 30 watts in a cooled IEF apparatus (Hoefer Scientific Instruments, San Francisco, CA). The plates were soaked in 100 ml of 10% CCl₃COOH, 5% 5-sulfosalicylic acid; stained with Coomassie Blue to detect IEF markers (Pharmacia Biotech Inc.); and autoradiographed. ³⁵S in the bands of more acidic (deamidated) and more basic (initial) X-DHFR species was quantified using a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).

S. cerevisiae nta1-carrying plasmids that expressed specific X-gal (Ub-X-gal) test proteins and either mouse NTAN1 (pSG61; see above) or S. cerevisiae Nta1p (pRB201 (27); a gift from Dr. R. T. Baker) were labeled with Tran³⁵S-label for 5 min, followed by a chase in the presence of cycloheximide, preparation of extracts, immunoprecipitation of X-gal, SDS-polyacrylamide gel electrophoresis, autoradiography, and quantitation essentially as described (27, 28, 29, 30, 58, 59).

RESULTS

Stewart et al. (39) have purified an amidohydrolase, termed Nt^N-amidase (also called protein-N-terminal asparagine amidohydrolase), from porcine liver that specifically deamidates N-terminal Asn in polypeptides, but does not deamidate free Asn. Stewart et al. (40) have also determined the amino acid sequence of several Nt^N-amidase fragments and used it to clone an ~1.3-kb porcine cDNA (Ntan1 cDNA) encoding this enzyme. To isolate a mouse Ntan1 cDNA, we screened a gt11-based mouse liver cDNA library with a porcine cDNA-derived probe and low-stringency hybridization (see ``Experimental Procedures''). The initially isolated 1560-bp mouse DNA insert contained a 114-bp stretch that was 85% identical to a region of the porcine Ntan1 mRNA; an even greater similarity between this region and its porcine counterpart was apparent at the level of deduced amino acid sequences. This 114-bp region was flanked by consensus sequences for mammalian intron/exon junctions (60), suggesting that the library's 1560-bp mouse DNA insert was derived from an unspliced or partially spliced precursor of Ntan1 mRNA. It is shown below that the 114-bp region of the 1560-bp insert is exon IX of the mouse Ntan1 gene.

Northern hybridization with the 1560-bp fragment as a probe detected an ~1.4-kb mRNA whose relative abundance in several mouse cell lines, including Friend erythroleukemia (MEL) cells, was significantly higher than in mouse liver (see below). Therefore, we screened a cDNA library from mouse MEL-C19 cells with the 1560-bp DNA fragment as a probe. This screen yielded a 1030-bp mouse DNA insert containing a 897-bp region that encoded a protein whose deduced amino acid sequence was 83% identical to the deduced sequence of porcine Nt^N-amidase, but lacked a region homologous to the sequence of the N-terminal 11 residues of the porcine enzyme. The method of 5-RACE-PCR (see ``Experimental Procedures'') was used to isolate the missing mouse cDNA sequence.

The resulting mouse Ntan1 cDNA clone contained a 930-bp ORF encoding a 310-residue (35 kDa) NTAN1 protein (Nt^N-amidase) with a calculated pI of 6.1, whose deduced amino acid sequence (Fig. 2A) was 88% identical to the deduced sequence of porcine Nt^N-amidase (40). The recently determined (deduced) sequence of human Nt^N-amidase is 92 and 91% identical to the deduced sequences of mouse and porcine Nt^N-amidases, respectively.⁶ The sequences of these Nt^N-amidases lack significant similarities to the sequences of other known proteins, including other amidotransferases. In particular, the sequence of the 35-kDa mouse NTAN1 protein was dissimilar to the sequence of the 52-kDa S. cerevisiae amidase (Nt-amidase) encoded by the NTA1 gene. (In contrast to Nt^N-amidase, the S. cerevisiae Nta1p Nt-amidase (a component of the yeast N-end rule pathway) deamidates N-terminal Asn or Gln (Fig. 1).) The deduced sequence of mouse NTAN1 (Fig. 2A) lacks motifs resembling membrane-spanning regions, signal sequences, or nuclear localization signals. Since Nt^N-amidase is likely to be a part of a multiprotein targeting complex analogous to the one observed in yeast (4), the absence of nuclear localization signal-like sequences in NTAN1 does not, by itself, preclude the possibility that Nt^N-amidase is located in both the cytosol and the nucleus.

A partially translated 208-bp region that begins at nucleotide 896 of the 930-bp mouse Ntan1 ORF (nucleotide 17031 in Fig. 2B) and encompasses the Ntan1 mRNA poly(A) addition site (Fig. 2B) is 98.6% identical to a 206-bp segment within the 3-flanking untranslated region of the mouse Il2 gene that encodes interleukin-2. The Il2-homologous region of Ntan1 contains the Ntan1 stop codon and the poly(A) addition site (Fig. 2B); neither of these functions is performed by the corresponding sequences of the Il2 locus. The near identity of sequences in these two regions of the mouse Ntan1 and Il2 loci and the absence of significant sequence similarities between the corresponding regions of human NTAN1 and IL2 (data not shown) suggest a recombination event in the mouse lineage that involved Ntan1 and Il2 and that occurred after separation of the mouse and human lineages.

A low-stringency Southern hybridization analysis of the BALB/c mouse kidney DNA digested with EcoRI, BamHI, or HindIII, using the 1560-bp fragment containing a putative 114-bp Ntan1 exon (see above) as a probe, revealed several hybridizing restriction fragments. A screening of a phage P1-based mouse genomic DNA library (derived from embryonic stem cells) for an insert containing the 114-bp Ntan1 exon was carried out (see ``Experimental Procedures''), yielding several P1 clones bearing ~100-kb mouse DNA inserts. Three of these inserts were found to contain the entire Ntan1 locus; one of them (P1 clone 1798) was used for nucleotide sequencing and exon mapping.

The mouse Ntan1 gene was found to span ~17 kb of DNA and to contain 10 exons whose length ranged from 54 to 177 bp (GenBank^TM/EMBL accession number U57691[GenBank]) (Figs. 2B and 3). The 79-bp exon I, encoding the first 26 residues of NTAN1, was located in an ~1.7-kb BamHI fragment, which also contained the putative promoter region of Ntan1 and the 5-untranslated leader sequence (Figs. 2B and 3). The 177-bp exon X encoded the last 59 residues of NTAN1 and was followed by the 138-bp 3-untranslated region containing the poly(A) addition site, which was also the boundary of a region homologous to the locus that encodes interleukin-2 (Figs. 2B and 3).

Among the intron/exon junctions, all of them except those of introns I and V contained the GT and AG consensus dinucleotides characteristic of mammalian nuclear pre-mRNA splice sites (60, 61). However, intron I contained TG (instead of consensus GT) and AC (instead of consensus AG) at the 5- and 3-splice sites, respectively, while intron V contained CG (instead of consensus AG) at the 3-splice site (Fig. 2B). A GT consensus dinucleotide was present 1 bp upstream of the predicted 5-splice site in intron I, while another consensus dinucleotide, AG, was located at the beginning of the coding region of exon II, 2 bp downstream of the expected splice site (Fig. 2B). If these consensus sequences were the actual splice junctions, the resulting DNA equivalent of the Ntan1 mRNA sequence in this region would have been CTGGAAAGA rather than CTGGAGGAAAGA actually present in the cloned Ntan1 cDNA. The latter sequence occurred not only in the Ntan1 cDNA derived from a MEL cell cDNA library, but also in the products of reverse transcription-PCR (see ``Experimental Procedures'') carried out using MEL cell mRNA and two independent sets of primers (data not shown). Moreover, the homologous region of porcine cDNA also contained ``nonconsensus'' splice junctions (40). We conclude that the usually strict constraints on the sequences of splice sites (61, 62) are bypassed at the sites that define intron V and especially intron I of the mouse Ntan1 gene.

To identify the promoter of mouse Ntan1, we first analyzed a region immediately upstream of exon I within the ~1.7-kb genomic BamHI fragment (Figs. 2B and 3). Nucleotide sequencing indicated the presence of multiple binding sites for the transcription factor Sp1 (63) immediately upstream of the Ntan1 ATG start codon. This area (Fig. 2B) resembled the GC-rich promoter regions of several weakly transcribed housekeeping genes (64); the similarities included the absence of TATA and CCAAT consensus sequences characteristic of many other eukaryotic promoters. However, given the ~1.4-kb size of Ntan1 mRNA (see below), the distance between the Ntan1 start codon and the poly(A) addition site (1068 bp), and the presumed size of the poly(A) tail (100-200 nucleotides), the Ntan1 promoter was expected to be present ~200 bp upstream of the start codon. The mapping of Ntan1 mRNA by primer extension (see below) also failed to indicate the presence of Ntan1 mRNAs whose 5-ends were located within 100 bp of the Ntan1 start codon.

Nucleotide sequencing revealed a putative TATA box consensus sequence 510 bp upstream of the Ntan1 start codon. A CCAAT consensus sequence was present 45 bp upstream of this TATA box (Fig. 2B). In a typical TATA/CCAAT-containing eukaryotic promoter, transcription starts 16-35 bp downstream of TATA (65, 66). Indeed, a primer extension analysis, using Ntan1 mRNA as a template and a synthetic oligonucleotide primer matching the 3-end of the sequence of exon I, mapped a set of major 5-ends of Ntan1 mRNA to the region ~18 to ~32 bp downstream of the above TATA box, with the 5-ends at positions 959 and 960 (Fig. 2B) being the most frequent ones (data not shown). On the assumption that the 5-ends of Ntan1 mRNA defined by primer extension were the actual transcription start sites (rather than sites resulting from a 5-processing of initial transcription products), these data indicated that the synthesis of Ntan1 pre-mRNA begins ~480 bp upstream of the Ntan1 start codon. Two other TATA box consensus sequences were present in the genomic DNA upstream of the sites corresponding to the mapped 5-ends of Ntan1 mRNAs (Fig. 2B), but these putative TATA boxes were not accompanied by CCAAT boxes.

The ~480-bp distance between Ntan1 transcription start sites and the Ntan1 start codon predicts the Ntan1 mRNA size of 1.7 kb (assuming the mean length of the Ntan1 mRNA poly(A) tail to be 0.15 kb), which is larger than the apparent size of Ntan1 mRNA in the Northern experiments (~1.4 kb; see below). This discrepancy may be due to an anomalous electrophoretic mobility of Ntan1 mRNA, to its cleavage during isolation, or, less likely, to the presence of an untranslated intron in the 5-region of Ntan1 pre-mRNA (no evidence for such an intron has been found).

In addition to the TATA and CCAAT sequences, the apparent promoter region ~0.5 kb upstream of the Ntan1 start codon (Fig. 2B) contains putative binding sites for ``general'' transcription factors such as CP-1, AP-1, E2A, ATF, and NF-I (63, 67, 68) as well as putative sites for other transcription factors such as c/EBP, AP-2, HNF-1, SDR, PEA3, Ets1, H-2DIIBP, NFuE1, NFuE2, PU.1, and HC3 (67, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79). The roles, if any, of these transcription factors in the expression of Ntan1 remain to determined.

The chromosomal location of Ntan1 was determined by interspecific backcross analysis using DNA of progeny derived from matings of ((C57BL/6J × M. spretus)F1 × C57BL/6J) mice. This interspecific backcross mapping panel has been typed over 2000 loci distributed among all of the mouse 19 autosomes as well as the X chromosome (53). C57BL/6J and M. spretus DNAs were digested with several restriction endonucleases and analyzed by Southern hybridization for informative restriction fragment length polymorphisms using a mouse Ntan1 genomic probe. The ~4-kb HindIII M. spretus-specific restriction fragment length polymorphism (see ``Experimental Procedures'') was used to follow the segregation of Ntan1 in backcross mice. The results of this mapping indicated that Ntan1 is located in the proximal region of mouse chromosome 16 and is linked to Prm1, Cebpd, and Igl (see ``Experimental Procedures'') (Fig. 4). Although 160 mice were analyzed for every marker to produce the segregation analysis data in Fig. 4, up to 185 mice were typed for some pairs of markers. Each locus was analyzed in pairwise combinations for recombination frequencies using these additional data. The ratios of the total number of mice exhibiting recombinant chromosomes to the total number of mice analyzed for each pair of loci and the most likely gene order were as follows: centromere-Prm1-(4:168)-Ntan1-(1:166)-Cebpd-(1:185)-Igl. The recombination frequencies (expressed as genetic distances in centimorgans ± S.E.) were as follows: Prm1-(2.4 ± 1.2)-Ntan1-(0.6 ± 0.6)-Cebpd-(0.5 ± 0.5)-Igl.

We have compared our interspecific map of chromosome 16 with a composite mouse linkage map that reports the map location of many uncloned mouse mutations (the mouse genome data base at the Jackson Laboratory, Bar Harbor, ME). Ntan1 is located in a region of the composite map that is relatively devoid of uncloned mouse mutations (data not shown). The proximal region of mouse chromosome 16 shares regions of homology with human chromosomes 16p, 8q, and 22q (summarized in Fig. 4) (80). Indeed, recent analyses, using fluorescent in situ hybridization and phage P1 clones containing human NTAN1 genomic DNA, localized the human NTAN1 gene to chromosome 16p (a region of homology to the chromosomal location of mouse Ntan1).⁷

Northern hybridization was used to compare the levels of Ntan1 mRNA among mouse tissues and cell lines. In addition to an Ntan1-specific probe, these comparisons included a probe for E2_14K, a mouse E2 enzyme whose sequence is similar to that of S. cerevisiae Ubc2p. The latter E2 enzyme (one of at least 11 distinct E2 enzymes in S. cerevisiae) is a component of the yeast N-end rule pathway (31, 32). We employed the rabbit E2_14K cDNA (52) to isolate a mouse cDNA that encodes E2_14K (see ``Experimental Procedures''). This cDNA was used as a probe in the Northern hybridizations (Fig. 5, B and D). The control probes in these analyses were cDNAs encoding mouse -actin and GAPDH (see ``Experimental Procedures'') (57). The level of Ntan1 mRNA in mouse tissues was ~100-fold lower than the level of -actin mRNA (data not shown).

A Northern blot containing approximately equal amounts of poly(A)⁺ RNA from various mouse tissues was hybridized with Ntan1 cDNA. This probe detected a diffuse ~1.4-kb band whose relative abundance (quantified using a PhosphorImager) varied at most 2-fold among different tissues (Fig. 5A). The testis-derived Ntan1 mRNA existed as two species: the minor one comigrated with the ~1.4-kb Ntan1 mRNA of the other tissues, while the major one had the apparent size of ~1.1 kb (Fig. 5A). It is unknown whether the testis-specific Ntan1 Northern pattern was entirely the result of RNA degradation during isolation. The two species of E2_14K mRNA (Fig. 5B) were apparently derived from two distinct primary transcripts (see ``Experimental Procedures''). In contrast to the relative constancy of Ntan1 mRNA levels among different tissues, the levels of E2_14K mRNA varied significantly, with the highest expression in skeletal muscle (Fig. 5, A and B). Consistent with this observation was the presence of several putative binding sites for MyoD, a muscle-specific transcription factor (81), in a region upstream of the start codon of the rat E2_14K ORF (GenBank^TM/EMBL accession number U04303[GenBank]) (data not shown); no putative MyoD-binding sites were detected in the Ntan1 promoter.

A Northern analysis of Ntan1 and E2_14K mRNAs was also carried out with poly(A)⁺ RNA prepared from Friend erythroleukemia (MEL-C19) and myoblast (C₂C₁₂) cell lines (Fig. 5, C and D). The relative amount of Ntan1 mRNA in MEL and C₂C₁₂ cells before differentiation was ~3- and ~20-fold higher, respectively, than the amount of Ntan1 mRNA in skeletal muscle (Fig. 5 and data not shown). Poly(A)⁺ RNA was also isolated from MEL cells that were induced to differentiate under conditions that resulted in 95% of the cells ceasing growth and producing hemoglobin (see ``Experimental Procedures''). The relative levels of Ntan1 mRNA were approximately the same in the growing, undifferentiated MEL cells and in their differentiated counterparts (Fig. 5C). Similar results were obtained with the mouse lymphoma EL4 cells that have been induced by a phorbol ester to differentiate and express interleukin-2 (data not shown). By contrast, the conversion of C<sub>2</sub>C<sub>12</sub> myoblasts (which contained the highest levels of Ntan1 mRNA among the mouse tissues and cell lines examined) into myotubes (see ``Experimental Procedures'') was accompanied by an ~4-fold decrease in the level of Ntan1 mRNA, whose relative amount in myotubes was slightly higher than in MEL cells, but significantly higher than in skeletal muscle (Fig. 5B).

The 88% similarity between the sequences of porcine Nt^N-amidase (40) and the mouse NTAN1 protein (Fig. 2A) strongly suggested that NTAN1 is also an Nt^N-amidase. To verify this conjecture, we asked whether extracts prepared from S. cerevisiae nta1 cells that lacked the endogenous yeast Nt-amidase (encoded by NTA1 (27)) but expressed mouse NTAN1 possess the expected Nt^N-amidase activity. The amidase assay employed ³⁵S-labeled derivatives of mouse DHFR that bore different N-terminal residues (see ``Experimental Procedures''). A ³⁵S-X-DHFR (X = Asn, Gln, or Asp) was incubated in an S. cerevisiae extract and subjected to IEF on a urea-containing polyacrylamide gel. This procedure fractionates a protein according to its pI (82). Since deamidation of Asn or Gln converts a neutral side chain into a negatively charged one, this modification should be detectable through a distinct shift in pI. Indeed, while the pI positions of Asn-DHFR and Gln-DHFR were indistinguishable, the pI position of otherwise identical Asp-DHFR that bore N-terminal Asp was shifted toward a lower pI (Fig. 6, lanes j-m; and data not shown).

S. cerevisiae nta1

nta1

nta1

S. cerevisiae nta1

The pI of ³⁵S-Asn-DHFR was not altered after its incubation with an extract from S. cerevisiae nta1 that lacked the NTA1-encoded yeast Nt-amidase (Fig. 6, lanes j and m). However, the incubation of ³⁵S-Asn-DHFR with an extract from otherwise identical nta1 cells that expressed mouse NTAN1 resulted in a pI shift of the DHFR substrate (Fig. 6, lanes g-i and p). In contrast to the yeast NTA1-encoded Nt-amidase, which deamidated either Asn-DHFR or (more slowly) Gln-DHFR (Fig. 6, lanes a-f and n-o), no deamidation of Gln-DHFR was observed with S. cerevisiae nta1 extracts containing mouse NTAN1 (Fig. 6, lane q), strongly suggesting that NTAN1 is specific for N-terminal Asn (see below for independent in vivo evidence bearing on this conclusion). In addition, the absence of charge modifications in the case of Gln-DHFR (but not in the case of Asn-DHFR) that has been incubated with NTAN1 (Fig. 6, lanes p and q) strongly suggested that the activity of NTAN1 is confined to N-terminal residues (no deamidation of internal Asn or Gln in Gln-DHFR). These findings confirmed and extended the observations made by Stewart et al. (39) with porcine Nt^N-amidase, using peptide-size amidase substrates and high pressure liquid chromatography fractionation of reaction products.

We asked whether the mouse Ntan1-encoded Nt^N-amidase (NTAN1) could function in the yeast N-end rule pathway. In particular, we wished to determine whether the confinement of the in vitro deamidating activity of mouse Nt^N-amidase (Fig. 7) to N-terminal Asn is retained under in vivo conditions. The test proteins were derivatives of E. coli gal. In eukaryotes, Ub fusions such as Ub-X-gal (X = a residue at the Ub-gal junction) are rapidly deubiquitylated by Ub-specific proteases to yield X-gal test proteins (2). In contrast to the function of Ub in protein degradation, the role of Ub in these engineered Ub fusions is to allow the in vivo generation of otherwise identical proteins bearing different N-terminal residues. Metabolic stabilities of X-gal proteins could be compared either indirectly (by determining the intracellular concentrations of X-gal proteins) (Fig. 7) or directly (in pulse-chase experiments) (Fig. 8). Previous work (27, 28, 29, 30, 31, 32) has shown that the enzymatic activity of an X-gal protein in yeast cells is a sensitive indicator of its metabolic stability.

nta1

nta1

nta1

Metabolic stabilities of Gln-gal, Asn-gal, and Arg-gal in S. cerevisiae nta1 that expressed mouse Nt^N-amidase.

An nta1 strain of S. cerevisiae that lacked the yeast NTA1-encoded Nt-amidase and expressed specific X-gal (Ub-X-gal) proteins such as Met-gal, Asp-gal, Asn-gal, or Gln-gal was transformed with plasmids expressing either S. cerevisiae Nta1p or mouse NTAN1, and the relative metabolic stabilities of X-gal test proteins were determined for each transformant by measuring the activity of gal (Fig. 7). In wild-type (NTA1) S. cerevisiae, the level of Met-gal (t1/2 > 20 h (6)) was much higher than the levels of Asp-gal, Asn-gal, and Gln-gal (t1/2 = ~3, ~3, and ~10 min, respectively (6)), with the level of Gln-gal being slightly higher than the levels of the two other normally short-lived X-gal proteins (Fig. 7). The same measurements in congenic nta1 cells showed that the level of Asp-gal (bearing a secondary destabilizing N-terminal residue) remained unchanged (in comparison to NTA1 cells), whereas the levels of Asn-gal and Gln-gal (bearing the tertiary destabilizing N-terminal residues) became nearly equal to the level of the long-lived Met-gal (Fig. 7; see also Ref. 6).

As expected, the expression of a plasmid-borne S. cerevisiae NTA1 gene in nta1 cells restored the low levels of Asn-gal and Gln-gal without altering the levels of Met-gal or Asp-gal (Fig. 7). However, the expression of NTAN1 in S. cerevisiae nta1 resulted in low levels of Asn-gal, but did not restore the low levels of Gln-gal, in contrast to the effect of the yeast Nta1p Nt-amidase (Fig. 7). These findings were confirmed by pulse-chase analyses of Asn-gal, Gln-gal, and Arg-gal degradation in yeast nta1 cells transformed with a plasmid expressing mouse NTAN1 (Fig. 8). We conclude that the in vitro selectivity of mouse Nt^N-amidase for N-terminal Asn was retained in vivo.

Although Nt-amidase activity (with Asn-DHFR as a substrate) in extracts from S. cerevisiae nta1 that expressed NTAN1 was significantly higher than the same activity in extracts from wild-type (Nta1p-containing) S. cerevisiae (data not shown), Asn-gal was more unstable in wild-type yeast than in congenic cells that lacked Nta1p and expressed mouse NTAN1 (Fig. 7). The apparently lower in vivo efficiency of the mouse Nt^N-amidase (in comparison to the yeast Nta1p Nt-amidase) in conferring a short half-life on a test protein such as Asn-gal may be caused by the likely absence of specific interactions between the mouse Nt^N-amidase and the targeting complex of the S. cerevisiae N-end rule pathway.

DISCUSSION

Stewart et al. (39, 40) reported the purification of a porcine Nt^N-amidase and the cloning of its ~1.3-kb cDNA. Described in the present paper is a further study of this enzyme, carried out with its mouse counterpart. We report the following results.

1) The mouse Ntan1 gene spans ~17 kb of genomic DNA and contains 10 exons ranging from 54 to 177 bp in length. The ~1.4-kb Ntan1 mRNA encodes a 310-residue (35 kDa) protein, NTAN1 or Nt^N-amidase.

2) The mapping of Ntan1, carried out using interspecific backcross analysis, located it in the proximal region of mouse chromosome 16, in an area devoid of uncloned mouse mutations.

3) The deduced amino acid sequence of mouse Nt^N-amidase lacks motifs that resemble membrane-spanning regions, signal sequences, or nuclear localization signals. The sequence of mouse Nt^N-amidase is 88 and 92% identical to the deduced sequences of porcine and human Nt^N-amidases, respectively. However, there are no significant sequence similarities between Nt^N-amidases and other known amidotransferases, including the NTA1-encoded Nt-amidase of the yeast S. cerevisiae. The 52-kDa Nta1p protein, a component of the S. cerevisiae N-end rule pathway, can deamidate N-terminal Asn or Gln (27), whereas the 35-kDa mammalian Nt^N-amidase can deamidate N-terminal Asn but not Gln (Fig. 1) (39).

4) The putative promoter region of Ntan1 is located ~500 bp upstream of the Ntan1 start codon. In addition to the TATA box and CCAAT consensus sequences, this region contains putative binding sites for many of the known transcription factors, including CP-1, AP-1, AP-2, E2A, c/EBP, SDR, PEA3, NFuE1, NFuE2, HNF-1, Ets1, H-2DIIBP, PU.1, and HC3.

5) Northern analyses of the ~1.4-kb Ntan1 mRNA showed it to be present in roughly equal amounts in all of the mouse tissues examined, including skeletal muscle, brain, and liver. The examined mouse cell lines expressed Ntan1 more strongly than did mouse tissues. While the in vitro differentiation of either Friend or lymphoma cells did not lead to a significant alteration in the amount of Ntan1 mRNA, its level was found to be decreased ~4-fold upon the conversion of C₂C₁₂ myoblasts into myotubes.

6) An isoelectric focusing assay was used to compare the in vitro enzymatic specificities of the mouse Nt^N-amidase and the yeast Nt-amidase. These experiments, using ³⁵S-X-DHFR test proteins and extracts from S. cerevisiae nta1 that either lacked or expressed mouse Nt^N-amidase, have shown that Nt^N-amidase deamidates N-terminal Asn but not Gln.

7) The expression of mouse NTAN1 in S. cerevisiae nta1 was found to restore the degradation of Asn-gal but not Gln-gal, indicating that mouse Nt^N-amidase can implement the Asn-specific subset of the yeast N-end rule and also showing that the in vitro selectivity of this enzyme for N-terminal Asn is retained in vivo.

The hierarchic structure of the N-end rule ``disperses'' the domains that recognize specific destabilizing N-terminal residues among several proteins such as Nt-amidase, R-transferase, and N-recognin (see the Introduction and Fig. 1). The regulatory possibilities of this arrangement are apparent, but physiological substrates of Nt-amidase and R-transferase remain unknown in both fungi and metazoans. Part of the difficulty in identifying these substrates stems from the fact that although a number of apparently cytosolic or nuclear proteins bear, at least initially, the N-terminal sequence Met-Asn or Met-Gln (39), the known Met-aminopeptidases are incapable of removing N-terminal Met if it is followed by an Asn or a Gln residue (4). In addition, at least the Met-Asn sequence is often acetylated at the N-terminal Met residue (83). It is unknown whether any of the resulting Ac-Met-Asn-containing N-terminal regions are processed in vivo to yield N-terminal Asn. Recently, certain sequences bearing N-terminal Met-Gln have been found to yield N-terminal Gln in vivo (84). The corresponding processing protease(s) remains to be identified. The isolation of Ntan1 and the resulting possibility of producing mouse ntan1 cells through targeted mutagenesis may provide additional routes to physiological substrates of Nt^N-amidase, for instance, through a search for proteins whose metabolic stabilities (or isoelectric points) differ between Ntan1 and ntan1 cells.

The possibility that certain cell types in a multicellular organism may lack Nt^N-amidase, yielding, in these cells, an N-end rule pathway that spares normally short-lived proteins bearing N-terminal Asn, has been made less likely by the finding that Ntan1 is expressed in all of the mouse tissues and cell lines examined. However, the Northern-based Ntan1 expression assays used thus far (Fig. 5) have been crude, their results being still compatible with the possibility of strong local differences in the levels of Ntan1 expression within specific tissues. Furthermore, if a binding site for Nt^N-amidase in a targeting complex of the mammalian N-end rule pathway overlaps with a binding site for the postulated Nt^Q-amidase, the competition between these enzymes for binding to the targeting complex and the likely presence of substrate channeling in the N-end rule pathway (27) may strongly influence the degradation of N-end rule substrates bearing N-terminal Asn versus those bearing N-terminal Gln. This hypothesis may account for the paradoxically slow degradation of Gln-gal (but not Asn-gal) in C₂C₁₂ myoblasts, which exhibited the highest level of Ntan1 mRNA among the tested mouse tissues and cell lines, but contained the postulated Nt^Q-amidase as well, as indicated by in vitro deamidation tests with extracts from C₂C₁₂ cells.⁵

The apparent absence of Ntan1 homologs in the mouse genome (those that can be detected by a low-stringency Southern hybridization) (data not shown) suggests that the postulated Nt^Q-amidase might be encoded by a differentially spliced mRNA derived from the Ntan1 gene. Indeed, although no differential splicing of mouse Ntan1 pre-mRNA has been detected thus far, the GenBank^TM/EMBL Data Bank was found to contain partial nucleotide sequences of apparent human NTAN1 cDNAs (accession numbers H27710[GenBank], H28425[GenBank], R10897[GenBank], N98532[GenBank], H44229[GenBank], N93132[GenBank], N80205[GenBank], and N69930[GenBank]), one of which (H28425), but not the others, lacks exon II of Ntan1 (data not shown). In addition, intron I (Figs. 2B and 3), which abuts exon II, contains noncanonical dinucleotide sequences at both 5- and 3-splice sites (see ``Results''). This circumstantial evidence for the hypothesis of a single gene encoding both Nt^N-amidase and Nt^Q-amidase remains to be verified directly.

The hierarchic organization of N-end rules, with their tertiary, secondary, and primary destabilizing residues (N-d^t, N-d^s, and N-d^p, respectively), is a feature that is more conserved in evolution than the Ub dependence of an N-end rule pathway or the identity of enzymatic reactions that mediate the hierarchy of destabilizing residues (4). For example, in a bacterium such as E. coli, which lacks the Ub system, the N-end rule has both N-d^s and N-d^p residues (it lacks N-d^t residues) (Fig. 1) (5, 85). The identities of N-d^s residues in E. coli (Arg and Lys) are different from those in eukaryotes (Asp and Glu in yeast; Asp, Glu, and Cys in mammals). Bacterial and eukaryotic enzymes that implement the coupling between N-d^s and N-d^p residues are also different: Leu/Phe-tRNA-protein transferase in E. coli and R-transferase in eukaryotes. Note, however, that bacterial Leu/Phe-tRNA-protein transferase and eukaryotic R-transferase catalyze reactions of the same type (conjugation of an amino acid to an N-terminal residue of a polypeptide) and use the same source of activated amino acid (aminoacyl-tRNA) (Fig. 1).

The apparent confinement of R-transferase to eukaryotes and of Leu/Phe-tRNA-protein transferase to prokaryotes (Fig. 1) suggests that secondary destabilizing residues were recruited late in the evolution of the N-end rule, after the divergence of prokaryotic and eukaryotic lineages. The lack of sequence similarity between the yeast Nt-amidase and the mammalian Nt^N-amidase and the more narrow specificity of the mammalian enzyme (Fig. 1) suggest that tertiary destabilizing residues (Asn and Gln) became a part of the N-end rule pathway much later yet, possibly after the divergence of metazoan and fungal lineages. If so, the N-end rule pathway may be an especially informative witness of evolution: the ancient origins of this proteolytic system, the simplicity and discreteness of changes in the rule books of N-end rules among different species, and the diversity of proteins that either produce or target the N-degron should facilitate phylogenetic deductions once the components of this pathway become characterized across a broad range of organisms.

Although the properties of the Ntan1-encoded mouse Nt^N-amidase are fully consistent with its being a component of the mouse N-end rule pathway and although mouse Nt^N-amidase was shown to be capable of implementing the Asn-specific subset of the yeast N-end rule in vivo, a rigorous test of the presumed function of this enzyme requires the production of mouse cell lines and/or whole animals that lack Ntan1. Work in this direction is under way.