Processing and Function of a Polyprotein Precursor of Two Mitochondrial Proteins in Neurospora crassa *

In Neurospora crassa , the mitochondrial arginine biosynthetic enzymes, N -acetylglutamate kinase (AGK) and N -acetyl- g -glutamyl-phosphate reductase (AGPR), are generated by processing of a 96-kDa cytosolic polyprotein precursor (pAGK-AGPR). The proximal kinase and distal reductase domains are separated by a short connector region. Substitutions of arginines at positions 2 2 and 2 3 upstream of the N terminus of the AGPR domain or replacement of threonine at position 1 3 in the mature AGPR domain revealed a second processing site at position 2 20. Substitution of arginine at position 2 22, in combination with changes at 2 2 and 2 3, prevented cleavage of the precursor and identified two proteolytic cleavage sites, Arg-Gly 2 Tyr-Leu-Thr at the N terminus of the AGPR domain and Arg-Gly-Tyr 2 Ser-Thr located 20 residues upstream. Inhibitors of metal-dependent peptidases blocked proteolytic cleavage at both sites. Amino acid residues required for proteolytic cleavage in the connector were identified, and processing was abol-ished by mutations changing these residues. The unprocessed AGK-AGPR fusion had both catalytic activities, including feedback inhibition of AGK, and complemented AGK 2 AGPR 2 mutants. These results indicate that cleavage of pAGK-AGPR is not required for functioning of these enzymes in the mitochondria. In eukaryotic organisms, arginine biosynthesis is compart-mentalized. In Neurospora crassa , the first six steps occur in the mitochondria Assays—N -Acetylglutamate kinase and N -acetyl- g -glu-tamyl-phosphate reductase of the

In eukaryotic organisms, arginine biosynthesis is compartmentalized. In Neurospora crassa, the first six steps occur in the mitochondria and the last two steps in the cytosol. Flux through the arginine biosynthetic pathway is regulated primarily by feedback inhibition of the enzymes that catalyze the first and second reactions. The second and third steps of the pathway are catalyzed by N-acetylglutamate kinase (AGK) 1 (EC 2.7.2.8) and N-acetyl-␥-glutamyl-phosphate reductase (AGPR) (EC 1.2.1.38) (1,2). These enzymes are produced from a polyprotein precursor (pAGK-AGPR), which is targeted to the mitochondria and processed into mature AGK and AGPR (3,4). The polyprotein consists of a mitochondrial targeting sequence followed by two protein domains, AGK and AGPR, separated by a connector region (Fig. 1A).
Most mitochondrial proteins are synthesized in the cytosol and targeted to the organelle by leader sequences at the N terminus of their precursors. Mitochondrial leader sequences are recognized by specific receptors on the mitochondrial outer membrane and translocated from the receptors to downstream components of the import machinery (5)(6)(7)(8). Removal of Nterminal targeting sequences in the matrix is performed by a mitochondrial processing peptidase (MPP), composed of two similar subunits, ␣-MPP and ␤-MPP; ␣-MPP is soluble in the matrix, and ␤-MPP is associated with the mitochondrial inner membrane (9 -12). As import and processing take place, proteins are folded into functional enzymes or assembled into functional multienzyme complexes. We previously showed that two proteins were obtained that comigrated with mature AGK and AGPR upon incubation of in vitro synthesized wild-type pAGK-AGPR with purified MPP (4). However, identification of the the precise cleavage site(s) in the connector region of the precursor remained to be determined.
Targeting sequences have the capability to form an amphipathic ␣-helix (13); however, defined sequences or structural motifs involved in proteolytic processing are not well understood (14 -17). The connector of pAGK-AGPR contains an internal processing sequence, which has some of the characteristics of a mitochondrial targeting sequence, although it is not predicted to form an amphipathic ␣-helix (4). Arginine residues at positions Ϫ2 or Ϫ3 and positions Ϫ10 or Ϫ11 relative to the first amino acid in the mature protein are often found in targeting sequences of mitochondrial precursor proteins and appear to form part of the not well defined motifs found at cleavage sites (15,16). Some similarities between the connector region of pAGK-AGPR and mitochondrial targeting sequences are apparent (Fig. 1B).
Several questions are addressed in this study. What are the sequences or structural motifs that specify the cleavage at the connector region of pAGK-AGPR? How many cleavage events are necessary to process the polyprotein precursor into two proteins? Is processing of the precursor into two independent proteins required for function in the mitochondria? Processing of pAGK-AGPR and its biological function were analyzed in vitro and in vivo. The roles of several amino acid residues as signals for processing were examined by introducing point mutations in the connector region of the precursor. Two sites for proteolytic processing were identified, and processing into two mature proteins was prevented by mutagenesis of these sites. Proteolytic cleavage of pAGK-AGPR in the connector region did not appear to be required for the activity of either enzyme or for feedback inhibition of AGK by arginine. Processing at the connector region of pAGK-AGPR is discussed in the context of putative advantages that targeting of fusion proteins may have versus the targeting of independent proteins.

MATERIALS AND METHODS
Strains and Growth Conditions-Escherichia coli strains DH5␣ and JM101 were used to propagate plasmid DNA. Strain GM48 was used to prepare nonmethylated DNA for digestion of methylation-sensitive restriction sites. E. coli RZ1032 and helper phage VCSM13 were used for the generation of single-stranded plasmid DNA for site-directed mutagenesis. Bacterial cultures were grown in LB medium or terrific broth (18) as specified. Neurospora wild-type strain LA1 (74A) and arg-6 strain LA358 (allele CD118) were obtained from R. H. Davis (19). Neurospora cultures were grown in Vogel's minimal medium N (20) or in minimal medium supplemented with 0.2 mg/ml arginine.
Construction of Plasmids-Plasmid constructs were derived from pUC19 (New England Biolabs) or pBluescript KSII(ϩ/Ϫ) (Stratagene). Constructs pRW5 and pJK2 have been previously described (4); pRW5 contains a copy of the wild-type arg-6 gene, which encodes pAGK-AGPR in an 8.0-kilobase pair SphI fragment cloned into pUC19; and pJK2 contains a full-length wild-type arg-6 cDNA in a 3.0-kilobase pair EcoRI-HindIII fragment cloned into pBluescript KSII(Ϫ). Plasmid pRW7 is a derivative of pRW5 that contains the hph gene on a 1.69-kb SalI-BamHI fragment. Plasmid pgH2 contains the arg-6 gene from pRW7 on a 4.1-kilobase pair HindIII fragment cloned into pUC19. Mutations in the connector region of the precursor were introduced by PCR using an 888-base pair Bsu361 fragment from plasmid pgH2 as a template or by site-directed mutagenesis using an 877-base pair ApaI-ClaI fragment from plasmid pJK2 (Fig. 1A, top diagram). A diagram of the introduced amino acid substitutions at the connector region of pAGK-AGPR is shown in Fig. 1B. The wild-type sequence of both the cDNA and genomic constructs is shown at the top. For simplicity, only the names of the cDNA constructs are indicated in the figure. Construct pcPG16 contains substitutions of the arginine pair at positions Ϫ3 and Ϫ2 relative to the N terminus of the AGPR domain to proline and glycine. Construct pcPG16 was used as the template for the introduction of additional mutations in several plasmid derivatives. Plasmid pcGTPG5 contains substitutions of arginine residues at positions Ϫ15 and Ϫ14 to glycine and threonine in a pcPG16 background. Constructs pcG22PG42, pcP22PG54, and pcG22PG10 contain a substitution of arginine at Ϫ22 to glycine or proline in a pcPG16 background; plasmid pcG22PG10 also has a stop codon toward the N-terminal region of the mature AGPR domain. Construct pcA13PG16 substitutes alanine for proline at position Ϫ13 in the pcPG16 background. Construct pcP3 substitutes proline for threonine at position ϩ3 in a wild-type background. Genomic constructs contain identical amino acid substitutions as the equivalent cDNA constructs and include a 56-base intron located toward the 3Ј-end of the distal AGPR coding region. The genomic constructs pgPG15-Hph, pgGTPG4-Hph, pgG22PG42-Hph, and pgP22PG54-Hph resulted from subcloning the 888-base pair Bsu361 fragment from pcPG16, pcGTPG5, pcG22PG42, or pcP22PG54 into pRW7.
Site-directed and PCR-mediated Mutagenesis-Oligonucleotides used for site-directed mutagenesis or PCR-mediated mutagenesis were synthesized using an Applied Biosystems model 391 PCR-mate DNA synthesizer (Table I). Site-directed mutagenesis was performed by the Kunkel procedure, modified for the use of single-stranded DNA derived from any plasmid (21). Single-stranded DNA was obtained by transformation of E. coli RZ1032 and infection of exponentially growing transformants with helper phage VCSM13 to a multiplicity of infection of 20. Annealing, extension by T4 DNA polymerase, and ligation were performed as suggested by Promega. PCR mutagenesis was performed by asymmetric PCR in a Perkin-Elmer thermocycler (22). Primary PCR reactions contained the template DNA pgH2 (ϳ10 ng), excess primer BsuAGK or IIB (100 pmol), limiting primer PG1 or PG2 (1 pmol), 1ϫ PCR buffer (20 mM Tris, pH 8.3, 25 mM KCl, 1.5 mM MgCl 2 , 0.05% Tween 20, 0.1 mg/ml gelatin), dNTPs (50 M), and Taq polymerase (2.5 units) in a final volume of 100 l. The PCR conditions for the denaturation, annealing, and extension reactions were 1 cycle at 94°C (2 min), 60°C (2 min), and 72°C (1 min) followed by 20 cycles at 94°C (1 min), 60°C (1 min), and 72°C (1 min) and 10 cycles at 94°C (1 min), 60°C (1 min), and 72°C (1.5 min). For the second PCR amplification, ϳ10 ng of each gel-purified first PCR product was used as a template. The outside primers, BsuAGK and IIB, were added, and PCR amplification was carried out using the same conditions as described above for the first amplification step. A low melting point agarose gel slice containing the second PCR product (ϳ660 ng) was resuspended in 80 l of 1ϫ Bsu361 buffer and digested with Bsu361 in a final volume of 100 l at 37°C for 60 min. The enzyme was inactivated by heating at 65°C for 10 min. For the ligation reaction, 50 l of the digestion mixture (ϳ330 ng) was added to 35 ng of Bsu361-linearized pgH2⌬Bsu361 in a final volume of 100 l of 1ϫ ligation buffer and incubated with 1 unit of T4 ligase at room temperature for 12 h (22). The ligation mixture was used to transform competent cells of E. coli DH5␣ (23). Constructs pgGTPG4, pcA13PG16, and pcP3 were also generated by asymmetric PCR using pgH2 as template DNA. The presence of the desired mutations was initially screened by digestion of DNA with SmaI (new site generated by the substitutions of Arg Ϫ2 and Arg Ϫ3 to Gly Ϫ2 and Pro Ϫ3 ), KpnI (new site introduced by the substitutions of Arg Ϫ14 and Arg Ϫ15 to Gly Ϫ14 and Thr Ϫ15 ), BstX1 (new site generated by the change Arg Ϫ22 to Gly Ϫ22 ), or HpaI (new site generated by the change Arg Ϫ22 to Pro Ϫ22 ) and subsequently by sequencing of the entire amplified regions.
In Vitro Transcription and Translation-To analyze processing in vitro, wild-type and mutated precursor proteins were synthesized by in vitro transcription and translation. Plasmid DNA was linearized with EcoRV, treated with proteinase K (50 g/ml) for 30 min at 37°C, and precipitated with 2 volumes of ethanol. DNA was resuspended in DEPC-treated distilled H 2 O and stored at Ϫ20°C. Proteinase-treated template was transcribed with T7 RNA polymerase as suggested by the manufacturer (Promega). Transcripts were visualized by electrophoresis using denaturing agarose gels (1.2% agarose containing 17% formaldehyde in MOPS buffer). Translation reactions in rabbit reticulocyte lysates were carried out in a final volume of 50 l as suggested by the manufacturer (Promega). In vitro transcribed RNA (2 l) was mixed with 1 l of RNasin (10 units/ml), 1 l of a mixture of amino acids (1 mM) except methionine, 5 l of [ 35 S]methionine (1,200 Ci/mmol, 10 mCi/ml), 6 l of H 2 O, and 35 l of reticulocyte lysate. The reaction was incubated at 30°C for 1 h. A 2.5-l aliquot was resolved by 7.5% SDS-PAGE and analyzed using a PhosphorImager (Molecular Dynamics). The rest of the sample was brought to a final concentration of 0.3 M sucrose and 0.05 mM methionine and stored at Ϫ80°C.
Isolation of Import-competent Mitochondria and in Vitro Import Assays-Mitochondria were isolated by differential centrifugation from N. crassa wild-type strain 74A grown for 12 h at 30°C in minimal medium (12). Import reactions were performed by incubating a suspension containing ϳ70 g of freshly isolated mitochondria in 100 l of import buffer (3% bovine serum albumin, 2. ) are indicated as AGK and AGPR, respectively. The internal processing sequence located in the connector region (ϳ200 amino acids) of the precursor is indicated as the internal processing sequence (IPS). The other portion of the connector region is indicated as the eukaryotic domain. Abbreviations for restriction endonucleases are as follows. B, Bsu361; A, ApaI; C, ClaI. B, diagram of internal processing sequence mutations; amino acid substitutions in each construct are indicated in boldface letters. mM) and EDTA (5 mM) were used (if indicated) as inhibitors of metaldependent peptidases (24,25). Prior to import, mitochondria were preincubated with the inhibitors at 25°C for 5 min in import buffer. After the addition of precursor to the reaction mixture, incubation was continued for 30 min. Postimport mitochondria were reisolated by centrifugation at 12,000 ϫ g for 12 min and treated (if indicated) with 50 or 100 g/ml of proteinase K for 30 min at 0°C to remove externally bound precursor. 35 S-Labeled import products were analyzed by electrophoresis in denaturing 7.5% polyacrylamide gels and fluorography.
Transformation of N. crassa-The genomic constructs pRW7, pgPG15-Hph, pgGTPG4-Hph, pgG22PG42-Hph, and pgP22PG54-Hph were introduced into the arg-6 recipient strain LA358 by transformation following the procedure of Vollmer and Yanofsky (26). Hygromycinresistant transformants were selected after 3-4 days of incubation at 30°C in regeneration agar overlaid on agar plates containing arginine (200 g/ml) and hygromycin (200 g/ml). Transformants were tested for arginine prototrophy by screening on plates without arginine but containing 100 g/ml hygromycin. Four isolates from each transformation were purified to homokaryons by four successive single conidial isolations. Purified transformants were analyzed by Southern and Western blots.
Crude Extracts and Immunoblots-Wild-type and ARG ϩ transformants were grown in 50 ml of minimal medium on a rotatory shaker at 30°C and harvested after 12 h by vacuum filtration. Mycelia (0.5 g, dry weight) were washed three or four times with deionized H 2 O, frozen in liquid nitrogen, and ground in liquid N 2 with a mortar and pestle to a fine powder. The powder was suspended in 1.5 volumes of boiling extraction buffer (3% SDS, 20 mM Tris-Cl, pH 8.0, 5 mM EDTA, 2 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride) and boiled for 5-7 min with vigorous mixing every 2.5 min. Cell debris was removed by centrifugation, and supernatants containing solubilized proteins were divided into 1-ml aliquots and frozen quickly on dry ice. Protein concentrations were determined with the Bio-Rad protein microassay. Solubilized proteins (50 g) were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Bio-Rad) by electroblotting at 4°C in transfer buffer (25 mM Tris-Cl, pH 8.3, 150 mM glycine, 20% methanol). The kinase (AGK) and reductase (AGPR) proteins were detected using anti-kinase or anti-reductase antiserum and an enhanced chemiluminescence system with anti-rabbit IgG-horseradish peroxidase as a secondary antibody (Amersham Pharmacia Biotech). Blots were exposed to film (Hyper-film type MP, Amersham Pharmacia Biotech) for time periods of 3-30 s.
N-terminal Sequencing-Mitochondria from transformants were purified by sucrose gradient centrifugation, and ϳ5-10 mg was sonicated three or four times for 10 s each in a Fisher sonic dismembrator, model 300, 35% maximal power, microtip probe in 1% TNET (50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, 1% Triton X-100) containing 2 mM phenylmethylsulfonyl fluoride. Mitochondrial membranes were removed by centrifugation at 15,000 ϫ g for 15 min at 4°C. Supernatants were brought to 400 l with 1% TNET, and a 40-l aliquot of undiluted anti-AGPR antiserum was added for each 5 mg of starting mitochondrial protein. Samples were incubated overnight with gentle rotation at 4°C. Immunocomplexes were precipitated with the equivalent of two volumes of Staphylococcus aureus extract as the source of protein A. Incubation was continued for 60 min on ice. Samples were centrifuged at 15,000 ϫ g for 1-2 min, and pellets were washed four times with 1 ml of TX/SDS (25 mM Tris-Cl, pH 7.0, 150 mM NaCl, 5 mM EDTA, 0.1% SDS, 0.05% Triton X-100) and once with 1 ml of TBS (25 mM Tris-Cl, pH 7.4, 100 mM NaCl, 1 mM KCl). Washed pellets were solubilized in 50 l of SDS/sample buffer without ␤-mercaptoethanol. Immunocomplexes were resolved by 7.5% SDS-PAGE and blotted onto polyvinylidene difluoride membranes (Bio-Rad) in CAPS transfer buffer (10 mM CAPS, 10% methanol, pH 11) at 4°C (27). Proteins on the membrane were stained with Coomassie blue R250, the band of interest was cut out of the membrane, and the amino terminus was sequenced by the UCLA Sequencing Facility.
Enzyme Assays-N-Acetylglutamate kinase and N-acetyl-␥-glutamyl-phosphate reductase activities of the uncleaved precursor were assayed using mitochondria purified by sucrose step gradient centrifugation (3). Acetylglutamate kinase was assayed by a radioactive procedure modified from Wolf and Weiss (28). Reactions contained 0.15 M Tris, pH 8.5, 60 mM MgCl 2 , 30 mM ATP, 3.75 mM [ 14 C]acetylglutamate, and 0.2 M NH 2 OH and were incubated for 80 min at 30°C. Feedback inhibition was measured in the presence of 18.75 mM arginine. The reactions were initiated by the addition of 200 l of mitochondrial extract and stopped by transferring to a boiling water bath. The reaction mixtures were acidified by adding 100 l of 0.5 N formic acid and centrifuged at 14,000 ϫ g to remove insoluble material. Supernatants were applied to an AG1-X8 (Bio-Rad) column (0.7 ϫ 6 cm) equilibrated with 0.1 N HCOOH and eluted with 0.1 N HCOOH. Glutamate was eluted in the first 10 ml followed by the reaction product, acetylglutamyl hydroxamate, in the next 10 ml; unreacted acetylglutamate was eluted with 1.0 N HCOOH. Fractions containing the product were dried in a vacuum dessicator, redissolved in 1 ml of 0.1 N HCOOH, and counted. Counts from a control reaction containing no ATP were subtracted from the values obtained in the test reactions. One unit of enzyme is defined as the amount required to produce 1 mol of product in 1 min at 30°C. The activity of acetylglutamate kinase is expressed in microunits and represents the average of three independent determinations.
Acetylglutamyl-phosphate reductase activity was assayed by following the increase in fluorescence as NADP ϩ was converted to NADPH (3). The reaction mixtures contained 0.1 M glycine, pH 9.3, 1.33 mM acetylglutamate 5-semialdehyde, 25 mM K 2 HPO 4 , pH 9.3, 0.67 mM NADP ϩ (freshly prepared), and 100, 200, or 300 l of mitochondrial extract in a final volume of 3 ml. The reaction was initiated by the addition of the substrate, acetylglutamate semialdehyde, and the increase in fluorescence was followed using a Gilson Spectra/Glo filter fluorometer (excitation filter, 330 -380 nm; emission filter, 430 -600 nm). The reactions were carried out at 25°C. The activity is expressed as the change in fluorescence/min and represents the average from three independent determinations.
Miscellaneous-Restriction enzymes were purchased from Promega and New England Biolabs. Taq polymerase was purchased from Pro-

Role of the Arginine Pairs at Positions
Ϫ2 and Ϫ3 and Positions Ϫ14 and Ϫ15 from the N Terminus of the AGPR Domain-The importance of an arginine residue at position Ϫ2 or Ϫ3 relative to the cleavage site of mitochondrial targeting sequences has been demonstrated (13)(14)(15)(16). Preliminary analysis indicated that replacement of arginine with glycine at position Ϫ2 relative to the N terminus of the AGPR domain (Arg Ϫ2 to Gly Ϫ2 ) did not prevent cleavage of the polyprotein in vitro. 2 A precursor containing the changes Arg Ϫ2 and Arg Ϫ3 to Gly Ϫ2 and Pro Ϫ3 (pcPG16) was used as a substrate for an in vitro import assay ( Fig. 2A). Postimport mitochondria were reisolated and treated or not treated with proteinase K to digest bound precursor. Processing of the wild-type precursor (lanes 1 and 4) resulted in two protein bands; the upper protein band corresponds to mature AGK (52 kDa), and the lower protein band corresponds to mature AGPR (37 kDa) (29). Processing of the mutant precursor derived from pcPG16 resulted in an AGK protein that comigrated with wild-type AGK, but the AGPR protein (AGPR*) appeared to have a higher molecular mass than wild-type AGPR ( Fig. 2A, lanes 2 and 5). This result suggested that processing at the N terminus of the AGPR domain, between amino acids Ϫ1 and ϩ1, was prevented by the amino acid substitutions and that one or more proteolytic cleavages took place at a location(s) upstream of the Ϫ1/ϩ1 site in the connector region of the precursor.
To investigate the role of the arginine residues at positions Ϫ14 and Ϫ15, mutations were introduced at these sites in a background that contained the previous changes at Ϫ2 and Ϫ3. Fig. 2A (lanes 3 and 6) shows that processing of the precursor derived from pcGTPG5, which contains the changes Arg Ϫ15 , Arg Ϫ14 to Gly Ϫ15 , Thr Ϫ14 and Arg Ϫ3 , Arg Ϫ2 to Pro Ϫ3 , Gly Ϫ2 resulted in an AGK protein band that comigrated with wildtype AGK (compare with lane 1) and an AGPR band with a higher molecular mass than wild-type AGPR. The larger AGPR band comigrated with the AGPR* observed in the processing of the precursor from pcPG16 (lanes 2 and 5). Thus, substitution of the arginine pair at Ϫ14 and Ϫ15 had no effect on processing at the second cleavage site.
Role of the Arginine Residue at Position Ϫ22 from the N Terminus of the AGPR Domain-Since mutations at the arginine pairs upstream of the Ϫ1/ϩ1 cleavage site identified a second processing site in the connector region, the role of the arginine at position Ϫ22 was examined. Arg Ϫ22 was chosen because it is in a context that resembles the cleavage site at Ϫ1/ϩ1. Analysis of the precursor derived from pcP22PG54 (Fig.  2B, lane 5), in which arginine at Ϫ22 was changed to Pro in a background where the arginine pair at Ϫ2 and Ϫ3 have been changed to Gly and Pro, shows that processing of the precursor was prevented by these substitutions, since no mature AGK or AGPR products were observed. The uncleaved precursor derived from pcP22PG54 was imported into mitochondria, since it was protected from proteinase K digestion. Some degradation of the precursor may have occurred in the matrix, suggested by the smear under the precursor band.
The role of arginine at position Ϫ22 can also be observed with construct pcG22PG10 (Fig. 2B, lane 4). The truncated precursor (T) was imported into mitochondria (protected from proteinase K) but was not cleaved (higher molecular mass than wild-type AGK). A full-length precursor containing the substitution at Arg Ϫ22 to Gly Ϫ22 (pcG22PG42; Fig. 2C, lanes 4 and 9) yielded similar results to those obtained with the precursor derived from pcP22PG54 (Fig. 2C, lanes 5 and 10); both precursors were imported (protected from proteinase K digestion) but were not processed in the mitochondria. However, some proteolytic degradation in the mitochondrial matrix was observed. These results indicate that proteolytic cleavage occurs at two different positions in the connector region of the precursor (Fig. 2D).
Cleavage of the Connector Region of the Precursor Is Inhibited by 1,10-Phenanthroline and EDTA-MPP cleaves mitochondrial targeting sequences during or after import of precursor proteins into the mitochondria. In N. crassa, removal of the targeting sequence by MPP can occur in more than one step (30). In some cases in yeast and mammals, processing of mitochondrial targeting sequences that are cleaved in two steps is carried out by two different enzymes; the first cleavage is performed by MPP, and the second cleavage is performed by a mitochondial intermediate peptidase that produces the mature protein (31)(32)(33). MPP and mitochondial intermediate peptidase are metal-dependent proteases inhibited by the chelators 1,10phenanthroline and EDTA (9, 10, 30), which do not affect the import of precursor proteins (24). Thus, it was of interest to investigate a possible role of a mitochondrial metal-dependent peptidase in the processing at the upstream cleavage site in pAGK-AGPR.
Import reactions with wild-type and mutated precursors were performed in the presence of inhibitors of metal-dependent processing peptidases (Fig. 3). To ensure that no precursor remained associated with the mitochondrial outer membrane and to evaluate the efficiency of proteolytic cleavage at the second site, postimport mitochondria were reisolated and treated with proteinase K. Wild-type precursor (Fig. 3, lanes 1  and 6) showed processing into two mature proteins, AGK and AGPR, in the absence of inhibitors (lane 1). Processing of the wild-type precursor was inhibited in the presence of 1,10-phenanthroline and EDTA (lane 6), as inferred from the presence of a protein band corresponding to unprocessed precursor and the almost complete absence of bands corresponding to mature AGK and AGPR. Processing of mutant precursors derived from pcPG16 (lanes 2 and 7) and pcGTPG5 (lanes 3 and 8) was also inhibited by 1,10-phenanthroline and EDTA, indicating that proteolytic cleavage at both processing sites is blocked by inhibitors of metal-dependent peptidases. Wild-type pAGK-AGPR is cleaved by purified MPP in the connector region to generate mature AGK and AGPR proteins (4). The importance of arginine at position Ϫ2 or Ϫ3 relative to the cleavage sites and the sensitivity to metal ion chelators suggest that MPP is likely to be responsible for the two-step proteolytic processing of the connector region of pAGK-AGPR.
Effect of Proline after an Arginine Pair on Processing-The arginine pair at positions Ϫ14 and Ϫ15 is located in a sequence motif that resembles the recognition site for MPP (14 -17) but did not appear to function as a signal for processing. Analysis of amino acid sequences flanking cleavage sites of several hormone and protein precursors revealed that most peptides that contain a Pro residue at position ϩ1 are not cleaved (34,35). To determine if the presence of a proline residue at position Ϫ13 disrupted a possible MPP cleavage site, the proline residue at position Ϫ13 was changed to an alanine in a background where the arginine residues at Ϫ2 and Ϫ3 were changed to Gly and Pro. Processing of the resulting precursor (pcA13PG16; Fig. 3, lane 4) resulted in three protein bands: unprocessed precursor and bands corresponding to wild-type AGK (compare with lane 1) and a larger AGPR that comigrated with AGPR* generated from pcPG16 and pcGTPG5 (lanes 2 and 3). Processing was completely inhibited in 1,10-phenanthroline/EDTA pretreated mitochondria (Fig. 3, lane 9). This result indicates that substitution of Pro for Ala in the wild-type sequence RRPAL is not sufficient to generate a cleavage site at RRA2AL. We conclude that arginine pairs by themselves are not sufficient for proteolytic cleavage in the connector region of pAGK-AGPR.

Role of Threonine in Recognition by MPP-A threonine res-
idue is often found 2 or 3 residues downstream from the cleavage site of mitochondrial targeting sequences. Interestingly, a threonine residue is also present 3 residues downstream from the cleavage site at the N terminus of the mature domain of the distal AGPR. To investigate the role of Thr ϩ3 in the AGPR FIG. 2. Import and processing of wild-type pAGK-AGPR and mutated derivatives by isolated mitochondria. A, import and processing of precursors derived from pcPG16 (Arg Ϫ2 and Arg Ϫ3 to Gly and Pro) and pcGTPG5 (Arg Ϫ2 and Arg Ϫ3 to Gly and Pro, and Arg Ϫ14 and Arg Ϫ15 to Thr and Gly). Precursor proteins were generated in a rabbit reticulocyte translation system in the presence of [ 35 S]methionine. Import reactions were carried out as indicated under "Materials and Methods." After import, mitochondria were reisolated and either not treated (lanes 1-3) or treated (lanes 4 -6) with 100 g/ml proteinase K (PTK). Samples were analyzed by SDS-PAGE and fluorography. Mutant AGPR protein is indicated as AGPR*. Plasmids carrying the mutated precursor genes are identified at the top of the lanes. B, import and processing of precursors from pcG22PG10 (Arg Ϫ2 and Arg Ϫ3 to Gly and Pro, and Arg Ϫ22 to Gly) and pcP22PG54 (Arg Ϫ2 and Arg Ϫ3 to Gly and Pro, and Arg Ϫ22 to Pro). Reisolated postimport mitochondria were treated with 50 g/ml proteinase K (lanes 1-5). Unprocessed precursor protein is indicated by P. Truncated precursor is indicated by T. Mutant AGPR protein is indicated as AGPR*. Plasmids carrying the mutant genes are identified at the top of the lanes. C, import and processing of precursors from pcG22PG42 and pcP22PG54. Postimport mitochondria were reisolated and either not treated (lanes 1-5) or treated (lanes 6 -10) with 100 g/ml proteinase K. Samples were analyzed by SDS-PAGE and fluorography. Names of the constructs examined are indicated at the top of each lane. Precursor protein is indicated by P. D, diagram of the processed and unprocessed products from wild-type AGK-AGPR and mutant derivatives. domain in processing of the connector region, this residue was changed to a proline in a wild-type background to generate the construct pcP3. In vitro processing of the resulting precursor resulted in three protein bands (Fig. 3, lane 5). The upper band corresponded to remaining unprocessed precursor; the two lower bands corresponded to wild-type AGK and to a larger sized AGPR (AGPR*) containing the N-terminal extension indicative of processing exclusively at the second cleavage site. Processing of the precursor was completely inhibited in 1,10phenanthroline/EDTA pretreated mitochondria (Fig. 3, lane  10). Thus, proteolytic cleavage at the N terminus of the distal AGPR was prevented by the threonine to proline substitution. These results suggest that threonine is a critical residue of the processing motif in the connector region of the precursor. The large fraction of unprocessed precursor (ϳ35% as measured by scanning densitometry) protected from proteinase K digestion suggests that processing at the second site is much less efficient than processing at the N terminus of AGPR.

Western Blot Analysis of in Vivo Expressed Constructs Containing Mutations in the Connector Region of the Precursor-To
analyze processing of pAGK-AGPR in vivo, strain LA358 (AGK Ϫ AGPR Ϫ ) was transformed with constructs containing different amino acid substitutions in the connector region of pAGK-AGPR. Fig. 4 shows the results of immunoblot analysis of transformants obtained with the construct pgGTPG4-Hph (Arg Ϫ15 , Arg Ϫ14 , Arg Ϫ3 , and Arg Ϫ2 to GTPG). This construct is the genomic equivalent to the cDNA construct pcGTPG5, which was processed in vivo at only the upstream second cleavage site. The immunoblot was probed with anti-AGK antisera (Fig.  4, panel A) and reprobed with anti-AGPR antisera (Fig. 4,  panel B). The recipient strain, LA358, has no detectable AGK or AGPR. AGK in the transformants (lanes T4 -T16) comigrated with wild-type AGK (74a). This result was consistent with the in vitro analysis of processing that showed generation of a wild-type AGK for the corresponding construct. The results for AGPR showed three types of transformants. One type expressed a wild-type AGPR (compare with 74A). A second type expressed only a larger sized AGPR (AGPR*), and a third type expressed both wild-type and mutant AGPR* (mAGPR*). These results were also observed with the construct pgPG15-Hph (Arg Ϫ3 and Arg Ϫ2 to Pro and Gly) (data not shown). The presence of the larger AGPR is consistent with the results observed in vitro, where processing only at the upstream cleavage site took place. The presence of wild-type AGPR can be explained by homologous recombination following transformation with the break point of the crossover located upstream of the site of the mutation. The presence of wild-type and mutant forms in the same isolate would result if the transformant is a heterokaryon containing both types of nuclei. These results confirm that processing of precursors containing amino acid substitutions at positions Ϫ2 and Ϫ3 from the N terminus of AGPR results in proteolytic cleavage at only the second site.
Processing of precursors derived from pgG22PG42-Hph (Arg Ϫ22 , Arg Ϫ3 , and Arg Ϫ2 to Gly Ϫ22 , Pro Ϫ3 , and Gly Ϫ2 ) and pgP22PG54-hph (Arg Ϫ22 , Arg Ϫ3 , and Arg Ϫ2 to Pro Ϫ22 , Pro Ϫ3 , and Gly Ϫ2 ), the genomic equivalents of pcG22PG42 and pcP22PG54, was analyzed by Western blotting; no mature AGK was observed (Fig. 5). These results are consistent with those observed in vitro and substantiate the conclusion that substitution at Arg Ϫ22 in combination with substitutions at Arg Ϫ2 and Arg Ϫ3 in the connector region of the precursor generates an uncleavable precursor.
Identification of the Second Cleavage Site-To identify the second site of proteolytic cleavage, the larger form of AGPR (AGPR*) was isolated from mitochondria of transformants pgPG15-Hph and pgGTPG4-Hph by immunoprecipitation, and the amino termini were sequenced (see "Materials and Methods"). The results are shown in Fig. 6. The variant AGPR* from both transformants had the same N-terminal sequence, indicating that cleavage occurred at the position between Tyr Ϫ20 and Ser Ϫ19 of the connector region. This resulted in an AGPR protein with a 20-amino acid N-terminal extension. Compari- FIG. 3. Effect of inhibitors of the mitochondrial processing peptidase on the import and processing of wild-type and mutant precursors. Isolated mitochondria were preincubated in import buffer without (lanes 1-5) or with (lanes 6 -10) 5 mM EDTA and 2.5 mM 1,10-phenanthroline (o-Phe) for 5 min at 25°C. 35 S-Labeled precursor was added, and incubation was continued for 30 min at 25°C. Mitochondria were reisolated and treated with proteinase K (50 g/ml) at 0°C for 30 min. To stop proteolysis, 1 mM phenylmethylsulfonyl fluoride was added, and the mixture was incubated for 5 min at 0°C. Postimport mitochondria were reisolated, and labeled proteins were visualized by SDS-PAGE and fluorography. Lanes 1 and 6, wild-type precursor; lanes 2 and 7, precursor from pcPG16; lanes 3 and 8, pcGTPG5; lanes 4 and 9, pcA13PG16; lanes 5 and 10, pcP3.
FIG. 4. Expression and processing of precursor derived from construct pgGTPG4-Hph. DNA from pcGTPG5 (Arg Ϫ2 and Arg Ϫ3 to Gly and Pro, and Arg Ϫ14 and Arg Ϫ15 to Thr and Gly) was subcloned into a vector containing the Hyg r selective marker (see "Materials and Methods"). The resulting construct, pgGTPG4-Hph, was transformed into strain LA358 (arg-6, allele CD118), which lacks AGK and AGPR proteins. A, Western blot analysis of initial heterokaryon transformants probed with anti-AGK antisera. Control lanes are wild-type strain, 74A (lane 1), and recipient strain, LA358 (lane 2). B, the same blot reprobed with anti-AGPR antisera. son of the sequence RG2YLT (first cleavage) and RGY2ST (second cleavage) reveals that the scissile bonds are in a sequence flanked by well conserved arginine and threonine residues (in boldface type). This suggests important roles for arginine and threonine as part of the cleavage site. Both residues appear to be critical for processing, since substitution of either of them results in misprocessing at the AGPR N terminus (see above).

Enzyme Activities and Feedback Inhibition in Uncleaved
AGK-AGPR Precursor-A question of major interest is whether processing of pAGK-AGPR into two mature proteins in the mitochondrial matrix is required for acetylglutamate kinase and acetylglutamyl-phosphate reductase to be active. Transformants expressing AGPR* with the N-terminal extension and those expressing an uncleaved precursor were able to grow in minimal medium at a rate comparable with that of wild type (not shown). This result indicated that processing was not essential for the functioning of these enzymes in the mitochondrial matrix. To obtain more direct evidence that proteolytic processing was not required, AGK and AGPR activities were measured in purified mitochondria from transformants expressing the uncleaved AGK-AGPR precursor. Transformants expressing uncleaved precursors derived from pgG22PG42-Hph (Arg Ϫ22 , Arg Ϫ3 , and Arg Ϫ2 to GPG) and pgP22PG54 (Arg Ϫ22 , Arg Ϫ3 , and Arg Ϫ2 to PPG) exhibited activities equal to or greater than wild type (Table II). In addition, feedback inhibition of AGK was not significantly affected by the lack of cleavage. These results show that the two-step proteolytic processing at the connector region of the pAGK-AGPR precursor to generate two mature proteins is not required for the biological activity of the two protein domains.

DISCUSSION
The 871-amino acid AGK-AGPR polyprotein precursor of N. crassa is organized as two protein domains separated by a 200-amino acid connector region with a 45-amino acid mitochondrial targeting sequence at the N terminus, which is cleaved into two mature proteins in the mitochondria (4). Proteolytic cleavage of leader sequences of mitochondrially targeted proteins has been shown to be performed in one or two steps by MPP (13)(14)(15)(16)(17) or in more than one step by two unrelated enzymes: MPP and a mitochondrial intermediate peptidase (31)(32)(33). MPP and mitochondial intermediate peptidase recognize and cleave different amino acid sequences. Processing of the internal processing sequence of the AGK-AGPR precursor involves removal of 22 residues upstream of the N terminus of the AGPR domain. Both cleavage events are inhibited by 1,10-phenanthroline and EDTA (this report) and take place upon incubation of in vitro synthesized precursor with purified MPP (4).
Kinetic studies of processing using oligopeptides with differ- FIG. 5. Expression and processing of precursor derived from constructs pgG22PG42-Hph and pgP22PG54-Hph. DNAs from constructs pgG22PG42-Hph (Arg Ϫ22 , Arg Ϫ3 , and Arg Ϫ2 to Gly, Pro, and Gly) and pgP22PG54-Hph (Arg Ϫ22 , Arg Ϫ3 , and Arg Ϫ2 to Pro, Pro, and Gly) were used to transform strain LA358 (arg-6, allele CD118), and transformants were selected by hygromycin resistance (see "Materials and Methods"). Crude extracts were prepared from initial hygromycin resistance isolates, and proteins were analyzed by immunoblot with anti-AGK antiserum. Lanes 1 and 8, molecular weight (MW) markers; lane 2, wild-type construct; lane 3, recipient strain LA358. Hyg r transformants with pgG22PG42-Hph and pgP22PG54-Hph are shown (as indicated at the top). Wild-type protein is indicated by AGK, and unprocessed precursor is indicated by P.
FIG. 6. Determination of second cleavage site in the connector region of the precursor by N-terminal sequencing of mutant AGPR proteins. Mutant proteins were purified by immunoprecipitation with anti-AGPR antiserum from crude extracts of the corresponding transformants and blotting to a polyvinylidene difluoride membrane. A second cleavage site was identified upstream of the N terminus of wild-type AGPR. A, wild-type sequence upstream of the N terminus of AGPR. B, mutant sequence from pgPG15-Hph (Arg Ϫ3 and Arg Ϫ2 to Pro and Gly). C, mutant from pgGTPG4-Hph (Arg Ϫ15 and Arg Ϫ14 to Gly and Thr, and Arg Ϫ3 Arg Ϫ2 to Pro and Gly). The positions of cleavages are indicated by arrows. Positively charged residues are indicated in boldface letters. Numbers under the amino acid indicate position relative to the N terminus of the AGPR domain. Italics indicate amino acid sequence data. ent amino acid sequences and lengths showed the necessity for at least 16 residues consisting of 11 residues upstream and 5 residues downstream from the cleavage site for effective hydrolysis by MPP (36). Analysis of amino acid sequences around the cleavage site of mitochondrial targeting peptides cleaved by MPP has revealed no consensus amino acids; however, several motifs conserved within various subgroups have been identified (14,15). In one subgroup, arginine residues were often observed at positions Ϫ2, Ϫ3, Ϫ10, and Ϫ11, relative to the scissile bond. In the connector region of pAGK-AGPR, arginine residues are present at positions Ϫ2, Ϫ3, Ϫ14, and Ϫ15 relative to the N terminus of mature AGPR. No motif resembling a cleavage site for mitochondial intermediate peptidase was identified. Although both MPP and mitochondial intermediate peptidase are metal-dependent proteases inhibited by 1,10phenanthroline and EDTA, analysis of the cleavage sites suggests that the two-step cleavage is carried out by MPP alone. Thus, maturation of pAGK-AGPR involves three proteolytic cleavage events: cleavage of the leader (targeting) peptide and two cleavages removing the internal processing sequence.
Cleavage at the connector region of the AGK-AGPR precursor occurs at sequences RGY2ST and RG2YLT, 20 amino acids apart. The results of in vitro import and processing of the mutant precursors support an important role for arginine at Ϫ2 or Ϫ3 and revealed the importance of a threonine at ϩ2 or ϩ3 relative to the cleavage site. A third sequence motif containing two arginines (RRPAL) occurs between the two cleavage sites. Substitution of alanine for proline, creating the sequence RRAAL, did not result in a new cleavage site.
The involvement of secondary structure in recognition by processing peptidases has been reported (34,35). Processing sites are believed to be in exposed and flexible regions of the precursors situated in, or immediately next to, ␤-turns or larger loops. The ␤-turn may constitute a key feature in the proteolytic processing reaction by providing a favorable conformation for optimal substrate-enzyme active site recognition. An average ␣-helix is 17 Å long and contains 11 residues, which corresponds to three turns. Individual ␤-sheets are, on average, 20 Å long, which corresponds to 6.5 residues (37). Most loops and turns occur at the surface and contain relatively polar residues. The most common type of loops link anti-parallel ␤-strands (␤-turns) or adjacent ␤-sheets (␤-hairpin). Predictions of local secondary structures in the connector region of the pAGK-AGPR precursor were performed using the combined algorithms of Chou-Fasman and Robson-Garnier (Mac Vector protein analysis package). The connector region of pAGK-AGPR is not predicted to form an amphipathic ␣-helix. However, local sheets and turns are predicted.
We examined possible correlations between the predicted perturbations in the local secondary structure of the connector region of the precursor and the presence or absence of proteolytic cleavage as shown by the processing assays. The wild-type precursor (Fig. 7A) shows three short stretches with the potential to form ␤-turns. ␤-Turns are apparent in the N terminus to C terminus direction around positions Ϫ40, Ϫ19, and Ϫ3. The formation of a ␤-sheet is apparent around position ϩ2 in the AGPR domain. Mutations in the arginine pair at positions Ϫ2 and Ϫ3 from the N terminus of the AGPR domain (pcPG16; Fig.  7B) decreased the predicted length of the ␤-turn at Ϫ3; the rest of the structure was not affected. Mutations at the arginine pairs at positions Ϫ2 and Ϫ3 and positions Ϫ14 and Ϫ15 from the C terminus of the AGPR domain (pcGTPG5; Fig. 7C) eliminated the turn predicted to be centered at Ϫ19. Substitution of Arg Ϫ22 with Gly Ϫ22 , combined with the substitutions at Ϫ2 and Ϫ3 (pcG22PG42; Fig. 7D) increased the length of the turn centered at Ϫ19. Substitution of Arg Ϫ22 with Pro Ϫ22 (pcP22PG54; Fig. 7E) shifted the turn centered at Ϫ19 toward the C terminus. Substitution of Pro Ϫ13 for Ala Ϫ13 in combination with changes at Ϫ2 and Ϫ3 resulted in the same structure predicted in Fig. 7B, where only changes at Ϫ2 and Ϫ3 were made (pcA13PG16; Fig. 7F). Substitution of Thr ϩ3 with a Pro ϩ3 in a wild-type background (pcP3; Fig. 7G) resulted in complete loss of the turn centered at Ϫ3 and the ␤-sheet at ϩ2. Therefore, specific amino acid replacements in the connector region of the precursor are predicted to affect the local secondary structure of the region. We found that in general, loss of proteolytic processing between the Ϫ1 and ϩ1 residues could be correlated with a decrease in the length or the complete loss of the ␤-turn centered at Ϫ3. However, loss of the ␤-turn centered at Ϫ19 did not affect processing at Ϫ20 (Fig. 7C). Loss of processing at position Ϫ20 is likely to result from the substitution of the arginine residue at Ϫ22. Thus, the predicted changes in the secondary structure caused by the amino acid substitutions consisted of variations in the length and position of ␤-turns, and some of these changes may be correlated with loss of proteolytic cleavage. Protein secondary structure as well as specific amino acid residues have been shown to be important for the processing of mammalian and plant precursors of mitochondrial proteins (38,39).  . 7. Predicted secondary structure of the connector region of wild-type and mutant precursors based on the combined algorithms of Chou-Fasman and Robson-Garnier. Predictions for ␣-helices (Hlx), ␤-sheets (Sht), or ␤-turns (Trn) starting at position Ϫ51 (T) to position ϩ11 (V), are indicated by shaded and black boxes. A window size of 7 was used. A, wild-type sequence; B, substitution of Arg Ϫ2 and Arg Ϫ3 by Gly and Pro; C, substitution of Arg Ϫ2 and Arg Ϫ3 by Gly and Pro, and Arg Ϫ14 and Arg Ϫ15 by Thr and Gly; D, substitution of Arg Ϫ2 and Arg Ϫ3 by Gly and Pro, and Arg Ϫ22 by Gly; E, substitution of Arg Ϫ2 and Arg Ϫ3 by Gly and Pro, and Arg Ϫ22 by Pro; F, substitution of Arg Ϫ2 and Arg Ϫ3 by Gly and Pro, and Arg Ϫ13 by Ala; G, substitution of Thr ϩ3 by Pro.
The biological advantage of targeting a fusion protein versus two independent proteins is not obvious; several possible reasons have been postulated (3,4). Facilitation of a multienzyme complex formation in the mitochondrial matrix for the channeling of labile intermediates is one explanation. In yeast, the His4 locus encodes a multifunctional protein with three different functional domains (40). Proteolytic processing of the His4 protein is not required for function, since the purified native protein contains the three activities.
AGK and AGPR activity assays using transformants expressing only an unprocessed pAGK-AGPR revealed that proteolytic cleavage in the connector region is not required for activity. In addition, the functions of AGK and AGPR in vivo were not affected as assessed by the ability of the unprocessed precursor to support growth of AGK Ϫ AGPR Ϫ mutants. We conclude that conformational changes that may be associated with the lack of proteolytic processing of the fused proteins did not have a dramatic effect on the function of the uncleaved enzymes. Moreover, the lack of processing did not affect feedback inhibition of AGK by arginine. However, whether interaction with other proteins in the mitochondrial matrix has been affected by the lack of processing remains to be studied.
The results shown here indicate that two proteolytic cleavages in the connector region of the precursor occur to release the two protein domains. Since cleavage in the connector region of the precursor takes place at positions Ϫ19 and Ϫ20 and positions Ϫ1 and ϩ1 from the N terminus of the mature AGPR, we propose that the enzyme responsible, possibly MPP, scans the connector region as the precursor reaches the mitochondrial matrix. The enzyme recognizes a first sequence for cleavage at Arg Ϫ22 , and it cleaves two residues toward the C terminus. The scanning continues as import of the precursor progresses. The enzyme skips the arginine pair at positions Ϫ15 and Ϫ14, probably due to the absence of a threonine residue in the recognition sequence. As scanning continues, the enzyme recognizes the arginine pair at positions Ϫ3 and Ϫ2 and cleaves 2 residues C-terminal. Since some unprocessed precursor is still observed after cleavage at the second site, cleavage at the N terminus of AGPR seems to be more efficient than the cleavage at the upstream position. At this point, folding of the processed distal domain into a functional protein begins, and the enzyme falls off the substrate.
What is the role of processing of the pAGK-AGPR precursor in the metabolism of arginine? It has been suggested that this precursor resulted from the fusion of two genes for independently targeted proteins. It has been hypothesized that this arrangement may result in a more efficient delivery of the proteins to the mitochondria or to stabilization of one or both proteins prior to their assembly into their mature functional forms (3,4,29). The increased activity of AGPR in transformants expressing the uncleaved polyprotein (Table II) suggests that possible effects on protein stability can be further enhanced by retaining the two enzymes as a polyfunctional protein. Because such transformants grow normally, the advantage of cleavage must be subtle, and its identification will require more extensive analysis of the properties of the polyprotein and independent enzymes.
Identification of the internal processing sites allowed identification of the precise C terminus of AGK. This confirmed the existence of a ϳ185-amino acid subdomain that is absent from prokaryotic homologs (4). A role for this subdomain may be related to the feedback inhibition properties of AGK in the mitochondrial matrix. Another possibility is that this subdomain plays a role in protein-protein interaction: mutations in arg-6 can affect the activity or feedback sensitivity of acetylglutamate synthase encoded by the unlinked arg-14 gene (28,41,42). Processing of the precursor may be needed for this interaction to occur. Kinetic analysis of the uncleaved precursor and examination of its interaction with other proteins may reveal new aspects of the role of proteolytic processing of the AGK-AGPR precursor on the metabolism of arginine.