The Human purH Gene Product, 5-Aminoimidazole-4-carboxamide Ribonucleotide Formyltransferase/IMP Cyclohydrolase

We report here the cloning and sequencing of the cDNA, purification, steady state kinetic analysis, and truncation mapping studies of the human 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/IMP cyclohydrolase (AICARFT/IMPCHase). These enzyme activities catalyze the penultimate and the final steps of de novo purine biosynthesis, respectively. In all species of both prokaryotes and eukaryotes studied, these two activities are present on a single bifunctional polypeptide encoded on the purH gene. The human purH cDNA is 1776 base pairs in length encoding for a 591-amino acid polypeptide (Mr 5 64,425). The human and avian purH cDNAs are 75 and 81% similar on the nucleotide and amino acid sequence level, respectively. The Km values for AICAR and (6R,6S)10-formyltetrahydrofolate are 16.8 mM 6 1.5 and 60.2 mM 6 5.0, respectively, for the cloned, purified human enzyme. A 10amino acid sequence within the COOH-terminal portion of human AICARFT/IMPCHase has some degree of homology to a previously noted “folate binding site.” Sitedirected mutagenesis studies indicate that this sequence plays no role in enzymatic activity. We have constructed truncation mutants which demonstrate that each of the two enzyme activities can be expressed independent of the other. IMPCHase and AICARFT activities are located within the NH2-terminal 223 and COOH-terminal 406 amino acids, respectively. The truncation mutant possessing AICARFT activity displays steady state kinetic parameters identical to those of the holoenzyme.

formyltetrahydrofolate. Interest in AICARFT stems in part from its potential as a chemotherapeutic target. Inhibitors of GARFT are currently in clinical trials as anti-neoplastic agents. Also, AICARFT inhibition is thought to be the origin of the anti-purine effects of anti-folates such as methotrexate whose primary target is dihydrofolate reductase.
In all organisms studied to date, AICARFT activity is accompanied by inosine monophosphate cyclohydrolase (IMPCHase, also known as inosinicase) located on the same polypeptide encoded by the purH gene. IMPCHase is the final step in the purine de novo pathway. The activities of the purine pathway in eukaryotes are frequently found on multifunctional proteins, whereas in bacteria, the enzymes are typically monofunctional proteins. Thus, AICARFT/IMPCHase is an exception in being bifunctional throughout evolution. A question which arises is whether there is some particular advantage which favors the bifunctional arrangement. For example, there might be kinetic "channeling" of the intermediate, formamidoimidazole ribonucleotide (FAICAR), between the penultimate and final catalytic centers. It is also possible that the enzyme could have a single binding site for the nearly complete purine ribonucleotide, which would be sequentially operated upon by two catalytic sites. Alternatively, there may be two functionally independent domains, similar to the situation with the other multifunctional proteins of the de novo purine pathway as it exists in eukaryotes.
AICARFT/IMPCHase from Bacillus subtilis (1), Salmonella typhimurium (2), Escherichia coli (3), and from chicken (4) have been cloned and expressed. Since there are no monofunctional proteins for sequence comparison, nothing is known about potential domain structure for AICARFT/IMPCHase from any species. Although GARFT and AICARFT/IMPCHase utilize the same cofactor and carry out very similar reactions, there has been no recognized sequence homology between them. This report describes the cloning, sequence, expression, and purification to homogeneity of the human AICARFT/IMP-CHase. Comparison of the human sequence with those of other species revealed a high degree of sequence conservation and homology. In addition, an amino acid sequence can be located which corresponds to a "folate binding consensus sequence" (5) previously noted in other 10-formyltetrahydrofolate binding enzymes such as GARFT. However, in contrast to other such sequences, this does not help to locate the catalytically active amino acid residues in AICARFT/IMPCHase and appears to be simply a common structural element. Truncation mutant studies identify a COOH-terminal fragment which possesses no IMPCHase activity but does possess AICARFT activity kinetically equivalent to the holoenzyme. An NH 2 -terminal fragment has also been identified which has IMPCHase activity but lacks AICARFT activity. These results suggest that AICARFT and IMPCHase activities exist as independent domains.

EXPERIMENTAL PROCEDURES
Cloning of the Human AICARFT/IMPCHase cDNA-The human AICARFT/IMPCHase (purH) cDNA was cloned from a human hepatoma ZapII cDNA library (Statagene, Inc.) utilizing described polymerase chain reaction (PCR) methods with some modifications (6). Some partial 3Ј nucleotide sequence information for the human purH cDNA had previously been obtained in the laboratories of Howard Zalkin and Jack Dixon, 2 who very kindly provided it to us. Based on this information, we designed human purH-specific oligonucleotide primers to isolate a 3Ј human purH cDNA fragment. Subsequently, PCR methods were used to produce fragments containing the 5Ј end (7). The fulllength human purH cDNA was assembled in the expression vector pET-14b (Novagen, Inc.) as shown in Fig. 2. 3 The resulting recombinant was termed pETHATNB-1800. A second full-length recombinant was constructed in the His-tag™ vector pET-23d (Novagen, Inc.), pETHAT-FLNN-1. The nucleotide sequence for each of the full-length clones was verified (8).
Expression of Human AICARFT/IMPCHase-The full-length human AICAR formyltransferase/IMP cyclohydrolase, point mutants, and truncation mutants were expressed and prepared from Escherichia coli BL21(DE3) (Novagen, Inc.) transformed with recombinant expression plasmids. E. coli transformants were grown at 30°C to an A 600 of 1.0. Isopropyl-␤-D-thiogalactopyranoside (Sigma) was added to a final concentration of 500 M, and incubation of the culture was continued for an additional 3 h at 30°C. The cells were subsequently harvested via centrifugation at 3500 ϫ g for 20 min at 4°C. Cell pellets were washed in ice-cold 0.85% NaCl and again pelleted via centrifugation. Pellets were either frozen at Ϫ70°C or used immediately.
Purification of the Full-length Human AICAR Formyltransferase/ IMP Cyclohydrolase-Purification of PurH was carried out at 4°C. Fresh or frozen cell pellets (approximately 12 g wet weight) were resuspended in 20 ml of ice-cold Buffer HB (20 mM Tris-Cl, pH 7.5, 150 mM NaCl, 50 mM KCl, 20% glycerol). Lysozyme was added to a final concentration of 10 mg/ml, and the suspension was incubated on ice for 20 min. Cells were subsequently sonicated on ice at 35% power for a total of 5 min using a Fisher Scientific Sonic Dismembrator model 300.
Cell lysates were clarified by centrifugation at 4°C, 39,000 ϫ g for 40 min in a Beckman Ti70 rotor. The clarified lysate was subsequently applied to a reactive red 120-agarose (Sigma) column (2.5 ϫ 12.0 cm) equilibrated in Buffer HB. The column was washed with 10 column volumes Buffer HB. Human AICARFT/IMPCHase was eluted using 20 mM Tris-Cl, pH 7.5, 500 mM NaCl, 50 mM KCl, 20% glycerol. Fractions were assayed for either AICAR formyltransferase or IMP cyclohydrolase activity. Peak activity fractions were pooled and were diluted with an equal volume of ice-cold Buffer HB. The enzyme was subsequently concentrated using an Amicon ultrafiltration cell (250 ml) with a Diaflo YM30 ultrafiltration membrane (Amicon, Inc.). A buffer exchange into Buffer HB was then performed, and the enzyme was concentrated to a final volume of 10 -15 ml. In order to further purify human AICARFT/ IMPCHase, the enzyme was desalted using a G25 coarse Sepharose (Pharmacia Biotech Inc.) column (1.5 ϫ 22.0 cm) equilibrated in Buffer B (20 mM Tris-Cl, pH 7.5, 20% glycerol). AICARFT/IMPCHase was immediately applied to an AICAR-Sepharose column (1.5 ϫ 5.5 cm) equilibrated in Buffer B. AICAR-Sepharose was prepared according to the method of Smith et al. (9) using commercially available CNBractivated Sepharose 4B (Pharmacia). The enzyme was eluted with 10 mM AICAR, 20 mM Tris-Cl, pH 7.5, 150 mM NaCl, 50 mM KCl, 20% glycerol. Fractions possessing IMPCHase activity were pooled and concentrated using Centricon 30 concentrators. A buffer exchange was performed into Buffer B. Typically, the enzyme was passed over the AICAR-Sepharose column a second time in order to further purify the AICARFT/IMPCHase. This second AICAR-Sepharose column elution was also followed by Centricon 30 enzyme concentration with a buffer exchange into Buffer HB. In order to determine if all AICAR had been removed, AICARFT activity assays were performed in the presence and absence of additional AICAR.
Purified protein and protein fractions were resolved throughout the purification by discontinuous SDS-polyacrylamide gel electrophoresis as described by Laemmli (10). Typical running and stacking gels used were 12% (w/v) and 4% (w/v), respectively. An SDS reducing sample buffer (62.5 mM Tris-Cl, pH 6.8, 10% glycerol, 2% SDS, 0.71 M ␤-mercaptoethanol, 0.0013 (w/v) bromphenol blue) was added to protein samples prior to electrophoresis. Samples were subsequently heated to 90°C for 5 min, loaded onto the SDS-polyacrylamide gel, and electrophoresed. Gels were stained with Coomassie Blue R250.
Mass Spectroscopy of Purified Human AICARFT/IMPCHase-Mass spectroscopy was performed at Yale University facilities. Samples were analyzed on a VG Quattro (Fisions Instruments, VG BioTech, Altrincham, Chesire, United Kingdom) triple quadrupole mass spectrophotometer equipped with an electrospray ion source (Analytica of Branford, Branford, CT). Data were collected and processed (smoothed, centered, and transformed) using the VG Mass Lynx software. Samples were dissolved in L-propanol and water (1:1 v/v) plus 0.1% formic acid and were subsequently injected into the electrospray ion source using a Rheodyne valve (model 7125) at a flow rate of 5 l/min. Nitrogen was used as a drying gas as well as to form the spray. The peak width at full width half-maxium for a singly charged ion was approximately 2 atomic mass units. Several scans each of 10-s duration were recorded in the multiple computer averaging mode.
FPLC Superdex 75 Gel Filtration Analysis of Human AICARFT/ IMPCHase-Individual experiments were conducted using human PurH purified from three independent preparations. Purified human AICARFT/IMPCHase was applied to an FLPC Superdex 75 column (Pharmacia) equilibrated in Buffer HB (described previously) and also in Buffer HB containing either 150 M 10-formyl-FH 4 , 100 M AICAR or both substrates at 4°C. When FPLC experiments were conducted in the presence of the substrates, the enzyme samples were preincubated with the substrate(s) for 30 min prior to loading the column. The column was run at a flow rate of 0.4 ml/min collecting 0.25-ml fractions. Fractions were subsequently assayed for either AICARFT or inosinicase activities. These experiments were conducted with each of the three human PurH preparations.
Molecular mass standard curves for the Superdex 75 column were calibrated using the elution profiles of the standard molecular mass markers (Pharmacia): albumin (67.0 kDa), ovalbumin (43.0 kDa), chymotrypsin (25.0 kDa), and ribonuclease A (13.7 kDa) monitored by ultraviolet absorbance at ϭ 280 nM. The column void volume was determined by the elution of dextran blue (2,000 kDa) from the column.
Enzyme Assays-Enzyme activity assays were carried out at room temperature. All assays were dependent on protein concentration and were initially linear with time. Protein concentration was determined by the Bradford protein assay (11). Required dye reagents were purchased from Bio-Rad. All assays were monitored using a Perkin-Elmer Lambda2 spectrophotometer employing the use of Perkin-Elmer PECSS computerized spectroscopy software.
Reaction mixtures (500 l ϭ final reaction volume) contained a final concentration of 33 mM Tris-Cl, pH 7.4, 25 mM KCl, 5 mM 2-mercaptoethanol, 0.1 mM (6R,6S)-10-formyltetrahydrofolate and 0.05 mM AICAR as described by Mueller and Benkovic (15). The reaction mixture was mixed prior to the addition of the human AICARFT/IMPCHase enzyme and subsequently mixed again. AICARFT activity was monitored by the formation of FH 4 at A 298 nm for 4 min according to the method of Black et al. (16) employing the use of a Perkin-Elmer Lambda2 spectrophotometer. The concentration of FH 4 formed was calculated by the difference in values between the extinction coefficients for 10-formyl-FH 4 and FH 4 , 19.7 ϫ 10 3 M Ϫ1 cm Ϫ1 (16).
IMP Cyclohydrolase Activity-FAICAR was synthesized according to previously published procedures (15,17). The concentration of FAICAR was determined using the extinction coefficient for FAICAR at A 248 nm and pH 7.4 (6.59 ϫ 10 3 M Ϫ1 ) (15). IMP cyclohydrolase enzyme activity reaction mixture consisted of 100 mM Tris-Cl, pH 7.4, and 0.1 mM FAICAR. Following mixing within the cuvette, the enzyme was added to a final reaction volume of 500 l. IMP cyclohydrolase activity was monitored by the appearance of IMP at A 248 nm for 4 min. The final concentration of IMP formed was determined by using the calculated value for the differences between the two extinction coefficients for IMP and FAICAR at A 248 nm and pH 7.4, 5.71 ϫ 10 3 M Ϫ1 (15,18).
Steady State Kinetic Analysis-The standard AICARFT assay was used for determining the steady state kinetic parameters for both the full-length human enzyme as well as the AICARFT activity-possessing truncation mutant. Experiments were performed varying concentrations of both AICAR as well as (6R,6S)-10-formyl-FH 4 . The data were fitted to the Michaelis-Menton equation using the EnzFitter program (Elsevier Biosoft) to calculate K m and V max values and their standard errors.
Construction of Truncation Mutants-Portions of the human AICARFT/IMPCHase cDNA were amplified via PCR from pETHATNB-1800 and were directionally subcloned and expressed in pET (Novagen, Inc.) vectors in order to identify possible functional activity domains. Three truncation mutants were generated: two amino-terminal truncation mutants expressing the first 223 and 230 amino acid residues of human PurH, respectively, and a carboxyl-terminal truncation mutant expressing the last 406 amino acid residues.
The first NH 2 -terminal AICARFT/IMPCHase cDNA truncation mutant, pETHAT-IMP, was constructed by PCR amplification of the first 669 nucleotides using the upstream and downstream primers 5Ј-CG-G(CCATGG)CTTCTCTATCAGCCTTATTTAG and 5Ј-GAGAT(GCG-GCCGC)GGGCTGCAGTGTGTACAGC, respectively (unique NcoI and NotI restriction endonuclease sites are in parentheses and the start codon underlined. A conservative amino acid change S2A, indicated in bold, was introduced to facilitate cloning and expression of human AICARFT/IMPCHase). The PCR product was resolved on a 1.2% agarose gel, gel-purified, and cloned into pCR™II (Invitrogen). The resulting plasmid, pTAIMP, was digested with NcoI and NotI, and the 669-bp 3Ј purH fragment was subsequently directionally subcloned into the NcoI and NotI polylinker sites of expression vector pET-23d creating plasmid pETHAT-IMP.
A second truncation mutant expressing the amino-terminal 230 amino acid residues was constructed. However, rather than cloning into the His-tag™ vector, pET-23d, we used vector pET-14b. Unique NcoI and BamHI restriction sites (designated in parentheses) were introduced into the 5Ј and 3Ј ends via oligonucleotide design. Truncation mutant pETHUIMP-230 was constructed using the upstream primer 5Ј-CGG(CCATGG)CTTCTCTATCAGCCTTATTTAG (described previously) and the downstream primer 5Ј-CCCAA(GGATCC)TCATA-GAACTGTGATGGGAAGC. Again, start and stop (antisense) codons are underlined.
The carboxyl-terminal truncation mutant, pETHATNB-1200, was generated using a similar strategy. The upstream and downstream primer combinations used include: 5Ј-GGA(CCATGG)CAATTTCAGAT-TATTTC/5Ј-CCCAT(GGATCC)TCAGTGGTGGAAGAGCCGAAGGTTC. Unique NcoI and BamHI restriction sites, indicated in parenthesis, were introduced into the 5Ј and 3Ј ends via oligonucleotide design. Start and stop (antisense orientation) codons are underlined. This recombinant was similarly constructed in vector pET-14b.
Partial Purification of Carboxyl-terminal Truncation Mutant Possessing AICARFT Activity-Plasmid DNA encoding PurH truncation mutant, pETHATNB-1200, was transformed into E. coli BL21(DE3). The cellular extract was prepared from isopropyl-␤-D-thiogalactopyranoside-induced transformants as described previously. The extract was loaded onto an AICAR-Sepharose column (1 ϫ 4 cm) equilibrated in Buffer B (described previously) and subsequently washed with 5 column volumes of Buffer B. The truncation mutant expressed from pETHATNB-1200 was eluted with 10 mM AICAR, 20 mM Tris-Cl, pH FIG. 2. Human AICARFT/IMPCHase cDNA clone. PCR was used to assemble the full-length 1776-bp human purH cDNA. Three overlapping partial human purH cDNA clones were obtained as presented under "Experimental Procedures." A, 5Ј purH cDNAs pHAT84 and pHAT4A were joined by PCR resulting in a 705-bp human 5Ј purH cDNA fragment. B, this 705-bp cDNA fragment was used in a subsequent PCR reaction with the 3Ј 1142-bp cDNA fragment. Unique NcoI and BamHI restriction sites (in brackets) were introduced via oligonucleotide design. A conservative amino acid replacement, S2A, was introduced for cloning purposes.
7.5, 150 mM NaCl, 50 mM KCl collecting 0.5-ml fractions. Fractions were analyzed by gel banding as well as AICARFT enzyme activity assays. Fractions determined to contain the truncation protein were subsequently pooled, concentrated using Centricon-10 filters (Amicon), and exchanged into Buffer HB.

RESULTS
Cloning and Sequence Analysis of the Human AICAR Formyltransferase/IMP Cyclohydrolase cDNA-The human AICARFT/IMPCHase cDNA clone, pETHATNB-1800, was obtained as described under "Experimental Procedures" and is shown in Fig. 2. The human purH cDNA nucleotide sequence has been submitted to GenBank™ with accession number U37436. The coding region of the human cDNA clone is 1776 nucleotides in length encoding for 591 amino acid residues (M r ϭ 64,425) and is presented in Fig. 3. Fig. 4 represents an amino acid sequence alignment of all reported full-length PurH sequences. Enzyme Purification and Kinetic Analysis-Human AICARFT/IMPCHase (M r ϭ 64,425) was purified essentially to homogeneity as described under "Experimental Procedures." The human AICARFT/IMPCHase was expressed at high levels using the pET-expression system and was purified over a series of four column steps. A typical purification is summarized in Table I. The initial reactive red 120-agarose column step typically resulted in a 2.5-6.2-fold purification. An apparent increase in total enzyme activity was consistently seen following the Amicon concentration step. Human AICARFT/IMPCHase . Darkly shaded regions represent amino acid residues that are strictly conserved. Lighter shaded areas represent those residues having a single base change on the nucleic acid level and are, therefore, noted as representing "mutationally" conservative amino acid changes. demonstrates higher activity in buffers containing 150 mM NaCl or in buffers containing 50 mM KCl perhaps as a result of stabilizing the enzyme conformation. The ionic strength of the enzyme buffer appears to have a greater effect on inosinicase activity. Removal of salt from the enzyme sample using a G25 coarse resin results in a 39 and 67% loss of AICARFT and IMPCHase specific activity, respectively. However, removal of salt is required for binding of the enzyme to the AICAR-Sepharose column. Two successive cycles of elution from an AICAR-Sepharose column were required to further purify the human AICARFT/IMPCHase to near homogeneity (Fig. 5). Since it has been well established that both enzyme activities reside on a single polypeptide, we typically followed inosinicase activity throughout the purification. A typical purification of the AICARFT/IMPCHase expressed from the cloned human purH cDNA resulted in a 22-and 39-fold purification of AICARFT and IMPCHase activities, respectively.
The steady state kinetic parameters for the purified human AICARFT/IMPCHase were determined (Table II). The AICARFT activity followed Michaelis-Menton kinetics for the determination of the K m values for both AICAR as well as 10-formyltetrahydrofolate. These kinetic data are consistent with values published for the native enzyme purified from MCF-7 human breast cancer cells (20) as well as the values obtained from human CCRF-CEM cells from two independent sources 4 (21).
We also conducted experiments to determine the K m for FAICAR. Consistent with results reported by Mueller and Benkovic (15) for the avian enzyme, our data for the human AICARFT/IMPCHase indicate that the K m for FAICAR is Ͻ1 M. The spectrophotometric assay for inosinicase activity is not sensitive enough to accurately measure values below 1 M.
Analysis of Human AICARFT/IMPCHase Aggregation State-The monomeric human AICARFT/IMPCHase has a calculated molecular mass of 64,425 daltons based upon the linear amino acid sequence. Electrospray mass spectroscopy data indicate a molecular mass of 64,450 daltons. We investigated possible formation of multimeric species using FPLC gel filtration. Human AICARFT/IMPCHase consistently eluted as a single activity peak (M r ϭ 65,630) whether or not either or both of its substrates were present at concentrations severalfold higher than their respective K m values, indicating that the protein exists as a monomer under the conditions used.
Localization of Separate AICARFT and IMPCHase Activity Domains-In order to determine whether the two activities of the human AICARFT/IMPCHase could be expressed on separate polypeptides, we generated several truncation mutants using PCR to amplify portions of the human AICARFT/IMP-CHase cDNA. The first truncation mutant, pETHAT-IMP, expresses the first 223 amino acids of the NH 2 -terminal portion of the full-length human clone in vector pET-23d. We also constructed an additional AICARFT/IMPCHase amino-terminal truncation mutant in vector pET-14b. The resulting plasmid, pETHUIMP-230, expresses the first 230 amino acid residues of the human AICARFT/IMPCHase coding region. A third truncation mutant was designed to express the final 406 amino acids of the COOH-terminal region of the human PurH protein. This COOH-terminal truncation mutant, encoded by plasmid pETHATNB-1200, overlaps with pETHAT-IMP by 111 nucleotides (37 amino acid residues) and with pETHUIMP-230 by 132 nucleotides (44 amino acid residues) (Fig. 6). Each of the truncation mutants as well as the wild type full-length human AICARFT/IMPCHase clone were transformed into E. coli. AICARFT and IMPCHase enzyme activity assays were carried out on extracts prepared from each of the transformants (Fig.  6). Extracts prepared from both pETHAT-IMP and pETHUIMP-230 transformants possessed IMPCHase activity, but lacked AICARFT enzyme activity. AICARFT enzyme activity was detected in extracts prepared from pETHATNB-1200 transformants and lacked IMPCHase activity. Both AICARFT and IMPCHase enzyme activities were detected in extracts prepared from transformants harboring the full-length human cDNA, pETHATNB-1800. Neither AICARFT nor IMPCHase activity was detected in E. coli harboring either vector pET-14b or pET-23d alone. These results demonstrate that the two enzyme activities are separable. In attempts to delineate the smallest peptides possessing either AICARFT or IMPCHase activity, truncation mutants expressing smaller portions of the AICARFT/IMPCHase carboxyl terminus or amino terminus were constructed (data not shown). SDS-polyacrylamide gel electrophoresis analysis of extracts prepared from these smaller constructs expressed in E. coli indicated that the peptides were expressed. However, enzyme activity assays indicated that these smaller polypeptides lacked both AICARFT as well as IMPCHase activity.
Kinetic Characterization of Carboxyl-terminal Truncation Mutant Possessing AICARFT Activity-The truncation mutant expressed from plasmid pETHATNB-1200 was partially purified as described under "Experimental Procedures." The steady state kinetic parameters of the partially purified truncation polypeptide were subsequently determined. Kinetic values were similar to those observed for the full-length human enzyme (Table II). The lower V max observed for the truncation mutant most likely results from impurities in the enzyme preparation. Refinement of the purification of this truncation mutant as well as evaluation of the enzyme mechanism and con- 4 O. Russello and G. P. Beardsley, unpublished results. ditions for crystallization are warranted. Analysis of a Putative N 10 -Formyltetrahydrofolate Binding Domain by Site-directed Mutagenesis-Previous reports had identified a putative 10-formyltetrahydrofolate binding consensus sequence among several enzymes that utilize 10-formyltetrahydrofolate (5). These data were based on an amino acid sequence alignment of several 10-formyl-FH 4 -requiring enzymes. A typical example is the relevant sequence of the human GARFT shown in Fig. 7. Active site irreversible inhibitors and mutational studies identified two amino acid residues, His 119 and Asp 144 , essential for activity in the E. coli enzyme (22,23). The corresponding residues in the human GARFT domain of the trifunctional human glycinamide ribonucleotide synthetase-aminoimidazole ribonucleotide synthetase-GARFT protein are His 915 and Asp 951 , respectively, and have been shown to be essential in the human enzyme as well (24). His 915 is located within the putative 10-formyltetrahydrofolate consensus sequence. Asp 951 is located downstream of this putative consensus sequence.
Amino acid sequence alignment of the human GARFT domain with the human AICARFT/IMPCHase indicate that this sequence is at least partially conserved. It was necessary to introduce a number of gaps in order to maximize the similarities between the human GARFT and AICARFT/IMPCHase amino acid sequences in the region under investigation. In addition, there are potential secondary structural differences between the human GARFT and AICARFT/IMPCHase sequences. GARFT possesses a proline residue which is not conserved within the human AICARFT/IMPCHase sequence. Moreover, this putative consensus sequence is conserved among the two known eukaryotic PurH sequences but not among any of the known prokaryotic PurH sequences. Nevertheless, these similarities warranted investigation.
Amino acid residues His 915 and Asp 951 , essential residues for GARFT activity, correspond to human AICARFT/IMPCHase residues His 469 and Asp 503 . It was noted that AICARFT/IMP-CHase Asp 501 was also conserved and due to its proximity to Asp 503 we also chose to study that residue as well. We investigated whether these residues were essential for either human AICARFT or IMPCHase activities using site directed mutagenesis. Three separate point mutations resulting in conservative amino acid replacements H469A, D501N, and D503N were each introduced into the human purH cDNA. Each of these three mutants, pETHAT-H469A, pETHAT-D501N, and pETHAT-D503N, as well as the wild type cDNA clone pETHATFLNN-1 and the vector pET-23d were expressed in E. coli. Protein expression was analyzed by SDS-polyacrylamide gel electrophoresis. Extracts prepared from each of the three mutants as well as the wild type purH cDNA clone indicated

DISCUSSION
The sequence of the human purH cDNA bears marked similarity to the previously reported avian sequence (4) with 75 and 81% identity on the nucleotide and amino acid sequence levels, respectively. The human sequence also demonstrates similarity to purH from prokaryotes, with amino acid identities ranging from 31 to 36% among the three bacterial sequences reported (1)(2)(3). While these sequences are similar, it should be noted that there appear to be distinct differences that may be characterized as being either more eukaryotic-like (human PurH amino acids Ile 276 -Leu 295 and residues Tyr 463 -Lys 524 ) or more prokaryotic-like (Gap intrajected between human PurH Phe 191 and Arg 192 ) (Fig. 4).
The steady state kinetic values for the AICARFT reaction using the human protein expressed in E. coli are comparable with those reported for native PurH derived from several human cell types. Thus, postsynthetic modification of the protein does not appear to be important for activity. The physical as well as the kinetic properties of the cloned human enzyme are likely to be similar to that of the native enzyme.
Enzyme cross-linking experiments for the avian enzyme conducted by Mueller and Benkovic (15) suggest that the avian AICARFT/IMPCHase may form a dimer. Under the conditions employed thus far, evidence indicates that the human enzyme exists as a monomer. It is of course possible that the human enzyme might form a dimeric or higher aggregate species under other conditions. We are continuing to investigate this possibility.
A question of major interest is whether the AICARFT and IMPCHase activities exist as separate functional domains or whether the two activities might share a common substrate binding site. To address this, we constructed a number of separate truncation mutants. Among these were found two NH 2 -terminal mutants, 223 and 230 amino acids in length, each of which lacked AICARFT activity but showed IMPCHase activity. IMPCHase activity can thus be localized within the first 223 amino acid residues of human AICARFT/IMPCHase. An additional truncation mutant expressing the COOH-terminal 406 amino acid residues possessed solely AICARFT activity. The degrees of overlap between each of the two NH 2terminal mutants and COOH-terminal truncation mutant are 37 and 44 amino acids. Construction of smaller amino-or carboxyl-terminal truncation mutants resulted in loss of IMP-CHase and AICARFT activity, respectively. Although it is possible that the small overlap region might include a common substrate binding site, this seems unlikely in view of its small size. It seems more likely that the bifunctional protein contains two separate functional domains. These might or might not be connected by a linking sequence. The next smallest NH 2 -terminal truncation mutant we have tested is 164 amino acids in length and shows no IMPCHase activity. The next smallest COOH-terminal truncation mutant is 364 amino acids in length and is likewise devoid of activity. Since the possibilities of unstable or misfolded proteins can not be ruled out, these mutants may not completely define the limits of the putative functional domains.
A further intriguing question is whether or not the two domains are functionally independent or whether the intermediate, FAICAR (Fig. 1), might be "channeled" to the IMPCHase domain. The COOH-terminal truncation mutant demonstrated independent function with kinetic properties highly similar to those of the holoenzyme. The available spectrophotometric assay for IMPCHase activity lacks the sensitivity necessary to permit comparison of activities between the truncation mutants and the holoenzyme. Thus, the question of potential channeling must await the development of a more sensitive assay.
We investigated a region of the human AICARFT/IMPCHase sequence corresponding to the putative 10-formyl-FH 4 binding domain of human GARFT and other 10-formyl-FH 4 -requiring enzymes using site-directed mutagenesis. Conservative amino acid replacements were introduced into the human purH cDNA by site-directed mutagenesis. These amino acid replacements had no effect on either AICARFT or IMPCHase enzyme activities for any of the mutant enzymes. While the residues His 915 and Asp 951 are essential for GARFT activity, the corresponding residues in the human AICARFT/IMPCHase are not essential. Therefore, this region may represent a common structural motif among these 10-formyl-FH 4 -requiring enzymes but is not involved in the single carbon transfer reaction nor in the binding of the folate substrate for AICARFT/IMPCHase. The crystal structure of E. coli GARFT shows that this region is near the folate binding site and may be important in the enzyme mechanism (25)(26)(27). Since no additional regions of similarity of the primary amino acid sequence between these two 10-formyl-FH 4 -requiring enzymes are apparent, the actual site of 10formyl-FH 4 binding for AICARFT/IMPCHase remains to be elucidated. Further investigation utilizing active site irrevers- ible inhibitors against either AICARFT activity or inosinicase activity will aid in the identification of residues essential for these activities.
Chemical modification studies by Szabados et al. (21) suggest that the IMPCHase activity of the human PurH requires an essential cysteine as well as an essential arginine residue. Amino acid sequence alignment of all known PurH sequences indicates that there are no cysteine residues conserved among all of the sequences (Fig. 4). There are, however, a total of 9 cysteine residues in the human AICARFT/IMPCHase, each of which is conserved in the avian AICARFT/IMPCHase sequence as well. Two of these cysteine residues, Cys 100 and Cys 145 , are localized within the amino-terminal truncation mutant expressing human IMPCHase activity. Based upon our truncation mutant studies, three arginine residues serve as potential candidates essential for IMPCHase activity. These arginine residues, Arg 63 , Arg 132 , and Arg 206 (numbered relative to the human sequence), are conserved among all known PurH sequences and are localized in large conserved "blocks" within the putative IMPCHase activity domain (Fig. 4). Residue Arg 206 is located within the 37 amino acid overlap between our two truncation mutants possessing either IMPCHase activity or AICARFT activity and may be a candidate for mutational studies. Other potential candidates include Arg 78 and Arg 170 which are conserved in all PurH sequences with the exception of Bacillus subtilis where there is a proximal arginine residue Expression of the cloned human purH cDNA in E. coli has permitted production of large quantities of human AICARFT/ IMPCHase. Further investigations of the structure and mechanism of this enzyme are now feasible.