Human 5-Aminoimidazole-4-carboxamide Ribonucleotide Transformylase/Inosine 5′-Monophosphate Cyclohydrolase

The bifunctional enzyme aminoimidazole carboxamide ribonucleotide transformylase/inosine monophosphate cyclohydrolase (ATIC) is responsible for catalysis of the last two steps in the de novo purine pathway. Gel filtration studies performed on human enzyme suggested that this enzyme is monomeric in solution. However, cross-linking studies performed on both yeast and avian ATIC indicated that this enzyme might be dimeric. To determine the oligomeric state of this protein in solution, we carried out sedimentation equilibrium analysis of ATIC over a broad concentration range. We find that ATIC participates in a monomer/dimer equilibrium with a dissociation constant of 240 ± 50 nm at 4 °C. To determine whether the presence of substrates affects the monomer/dimer equilibrium, further ultracentrifugation studies were performed. These showed that the equilibrium is only significantly shifted in the presence of both AICAR and a folate analog, resulting in a 10-fold reduction in the dissociation constant. The enzyme concentration dependence on each of the catalytic activities was studied in steady state kinetic experiments. These indicated that the transformylase activity requires dimerization whereas the cyclohydrolase activity only slightly prefers the dimeric form over the monomeric form.

The bifunctional enzyme aminoimidazole carboxamide ribonucleotide transformylase/inosine monophosphate cyclohydrolase (ATIC) is responsible for catalysis of the last two steps in the de novo purine pathway. Gel filtration studies performed on human enzyme suggested that this enzyme is monomeric in solution. However, cross-linking studies performed on both yeast and avian ATIC indicated that this enzyme might be dimeric. To determine the oligomeric state of this protein in solution, we carried out sedimentation equilibrium analysis of ATIC over a broad concentration range. We find that ATIC participates in a monomer/dimer equilibrium with a dissociation constant of 240 ؎ 50 nM at 4°C. To determine whether the presence of substrates affects the monomer/dimer equilibrium, further ultracentrifugation studies were performed. These showed that the equilibrium is only significantly shifted in the presence of both AICAR and a folate analog, resulting in a 10-fold reduction in the dissociation constant. The enzyme concentration dependence on each of the catalytic activities was studied in steady state kinetic experiments. These indicated that the transformylase activity requires dimerization whereas the cyclohydrolase activity only slightly prefers the dimeric form over the monomeric form.
Interest in ATIC partly comes from the fact that it is a potential chemotherapeutic target. AICAR TFase, along with the third enzymatic activity in the pathway, glycinamide ribonucleotide transformylase (TFase), utilizes the same reduced folate substrate making these two enzymes potential targets for anti-folate drugs. The anti-folate (6S/R)5,10-dideaza-5,6,7,8,-tetrahydrofolate (DDATHF) has been shown to be a potent antiproliferative agent, blocking de novo purine synthesis by inhibition of glycinamide ribonucleotide TFase and possibly AICAR TFase (1)(2)(3)(4). Inhibition of AICAR TFase by metabolites of the folate analog methotrexate has been implicated in the mechanism of controlling inflammation due to arthritis and other inflammatory diseases (5)(6)(7).
The human form of ATIC (hATIC) was first cloned in our laboratory from a hepatoma cDNA library (8). Sugita et al. (9) later found an additional form of hATIC cloned from a human placenta cDNA library. This isoform differs from our original form in the six N-terminal amino acids, as well as in a substitution of glycine for aspartate at position 165 (9). We will refer to these isoforms as hATIC-a and hATIC-b, respectively. Because results from Sugita et al. (9) suggest that hATIC-b is more prevalent in human cells, this isoform of hATIC was used in all experiments described here.
Previously reported experiments attempting to determine the oligomeric state of ATIC have yielded contradictory results. Enzyme cross-linking experiments using both yeast isoforms and the avian form of the enzyme suggest that ATIC may form a dimer (10,11). Gel filtration experiments using recombinant human enzyme conducted by Rayl et al. (8) suggested that ATIC exists as a monomer under the conditions tested (8). These apparently conflicting conclusions may actually both be valid, because the results depend greatly on the experimental conditions used to study the oligomeric properties of a protein.
The questions of whether ATIC exists as a monomer or some other higher oligomeric species and which oligomeric form of the enzyme is the active species need to be addressed because of their importance for further understanding the structurefunction relationships of this enzyme. This information will be critically important for postulating catalytic mechanisms for each activity and in the structure-based rational design of inhibitors targeting this enzyme.
In this report we demonstrate that ATIC exists in a monomer/dimer equilibrium and that this equilibrium is perturbed when both a folate analog and AICAR substrate are present but is not significantly shifted when each substrate is added individually. Evidence in support of these conclusions comes primarily from sedimentation equilibrium analytical ultracentrifugation studies of hATIC with and without substrate. We also show that for the IMPCHase enzymatic activity, both oligomeric forms possess catalytic activity with the dimeric form being more active. For the AICAR TFase enzymatic activity, the dimer appears to be the active form with the monomer possessing little or no catalytic activity.

EXPERIMENTAL PROCEDURES
Cloning of hATIC-b Expression Vector from hATIC-a-To clone the hATIC-b isoform of hATIC, PCR was used to change the first six amino acids, as well as the glycine at position 164, in the original clone hATIC-a using pETHATNB-1800 (8) as the initial PCR template. The primers used in the initial PCR reaction to change the first six amino acids were as follows: ATIC-a/b sense, 5Ј-GAGTTAAGCTAGCATG-GCTCCCGGCCAGCTCGCCTTATTTAGTGTCTCTGAC and ATIC antisense, 5Ј-CCCATGGATCCTCAGTGGTGGAAGAGCCGAAGGTTC. Unique NheI and BamHI sites are underlined, and the sequence changes are indicated in bold. The PCR product was resolved on a 1% agarose gel, purified using the QIAquick gel extraction kit (Qiagen), and double-digested with NheI and BamHI (New England BioLabs, Inc.). The resulting fragment was directionally cloned into the NheI and BamHI polylinker sites of expression vector pET28a(ϩ) (Novagen, Inc.) using T4 DNA ligase (New England BioLabs, Inc.). Selection of these restriction sites incorporated an N-terminal histidine⅐tag onto the clone to facilitate purification. This new clone was named pEThybrid-a/b.
Site-directed mutagenesis of G165D 2 using the method of overlap extension (12) was then carried out using pEThybrid-a/b as the template. Two PCR reactions were carried out using the following primers: ATIC-b sense, 5Ј-GAGTTAAGCTAGCATGGCTCCCGGCCAGCTC and G165D reverse, 5Ј-CTCCAAGGAGGTGTCCTTACTCTCGGAGC in the first reaction, and G165D forward, 5Ј-CTCCGAGAGTAAGGACAC-CTCCTTGGAG and ATIC antisense (shown above) in the second reaction. The NheI site is underlined in the ATIC-b sense primer, and the G165D mutation is indicated in bold. These PCR products were resolved on a 1% agarose gel and purified as above. These products were then used in a second PCR reaction acting as both primer and template along with the ATIC-b sense and ATIC antisense primers to generate the full-length ATIC-b cDNA containing NheI and BamHI restriction sites. This product was inserted into pET28a(ϩ) as described above creating plasmid pETATIC-b. The sequence of this new clone was verified by automated sequencing at the Howard Hughes Medical Institute Biopolymer/W. M. Keck Foundation Biotechnology Resource Laboratory at Yale University.
Human ATIC Expression and Purification-Plasmid pETATIC-b was transformed into Escherichia coli Tuner™(DE3) (Novagen, Inc.) by heat shock. E. coli transformants were grown in 2YT bacterial media (13) with 50 g/ml kanamycin (International Biotechnologies, Inc.) at 37°C to an A 600 value between 0.6 and 0.8. These cultures were then induced with 400 M isopropyl-␤-D-thiogalactopyranoside (American Bioanalytical) and grown at 30°C for 3 h. The cells were harvested by centrifugation, washed with cold 0.85% NaCl, and either used immediately or stored at Ϫ80°C.
Cell pellets were dissolved in cold lysis buffer (20 mM NaH 2 PO 4 , pH 8.0, and 100 mM NaCl). The cell suspension was lysed by sonication on a Fisher sonic dismembraner, model 300. The lysate was cleared by ultracentrifugation at 4°C, 15,000 ϫ g for 30 min in a Beckman Ti45 rotor. The supernatant solution containing soluble ATIC protein was poured off and placed on ice.
The immobilized metal affinity chromatography purification was performed on a BioCAD ® Sprint chromatography work station (PerkinElmer Life Sciences) using a column containing POROS metal chelate media (PerkinElmer Life Sciences). The column was charged with 5 column volumes (CV) of 0.1 M NiSO 4 ⅐6H 2 O, washed with 10 CV of water, and equilibrated with 10 CV of 99.9% lysis buffer and 0.1% elution buffer (20 mM NaH 2 PO 4 , pH 8.0, 100 mM NaCl, 250 mM imidazole). The column was then loaded with 10 -20 ml of cell lysate, washed with 99.9% lysis buffer and 0.1% elution buffer for 5 CV, and gradientwashed for 20 CV starting with 99.9% lysis buffer and 0.1% elution buffer and ending with 80.0% lysis buffer and 20.0% elution buffer. ATIC was eluted with 5 CV of 100% elution buffer collecting 2-ml fractions. The flow rate was 10 ml/min throughout the purification. Fractions were then assayed for either IMPCHase or AICAR TFase activity as described below. Those fractions containing activity were pooled and concentrated to 10 -15 ml using a 250-ml Amicon ultrafiltration cell with a Diaflo YM30 ultrafiltration membrane (Amicon, Inc.). The protein solution was loaded onto a HiLoad 16/60 Superdex 200-pg gel filtration column (Amersham Pharmacia Biotech) equilibrated and washed with HB buffer (20 mM Tris-Cl, pH 7.5, 150 mM NaCl, 50 mM KCl, 5 mM EDTA, 5 mM dithiothreitol) at a rate of 1 ml/min for 150 ml on a fast performance liquid chromatography system (Amersham Pharmacia Biotech). Fractions (1 ml) containing activity were pooled and stored at 4°C in HB buffer until needed.
Enzyme Activity Assays-All enzyme activity assays were performed on a PerkinElmer Life Sciences UV-visible Lambda2 spectrophotometer with PECSS data acquisition software unless noted otherwise. All reaction mixtures were 500 l in volume and performed in a 1-cm pathlength quartz cuvette at room temperature unless noted otherwise.
AICAR TFase activity was followed by monitoring the formation of tetrahydrofolate at A 298 based on a previously reported method (11,14). Each reaction mixture contained a final concentration of 66 mM Tris-Cl, pH 7.4, 100 M 10-f-FH 4 , 50 M AICAR, and 50 mM KCl unless noted otherwise. Each assay was initiated by rapid mixing of the substrates with the assay solution in the cuvette. The 10-f-FH 4 was synthesized as described previously (8,15,16) from (6S)5-formyl-5,6,7,8-tetrahydrofolic acid (leucovorin) purchased from Schircks Laboratories. AICAR was purchased from Sigma.
IMPCHase activity was followed by monitoring the formation of IMP at A 248 based on a previously reported method (11). Each reaction mixture contained a final concentration of 100 mM Tris-Cl, pH 7.4, and 0.1 mM FAICAR unless noted otherwise. Each assay was initiated by rapid mixing of the substrate with the assay solution in the cuvette. FAICAR with synthesized from AICAR (Sigma) according to previously published procedures (11,17).
Sedimentation Equilibrium of hATIC-b-A Beckman Optima TM XL-I ultracentrifuge was used to determine the equilibrium distribution of hATIC-b in HB buffer using the Rayleigh interference optics to visualize the protein. The interference experiments were carried out using external loading two-sector cells with sapphire windows. Data were collected at 4°C at three initial concentrations of 3.5, 8.7, and 10.7 M for hATIC-b at rotor speeds of 10,800, 13,300, and 16,000 rpm. For all experiments, water-water blanks were obtained at each speed prior to running the solutions of interest without disassembly of the cells between runs. Both the sample and buffer volumes were 125 l.
The effect of substrates/inhibitors upon the equilibrium distribution was also investigated. subtracting the water-water blanks from the interference data. The attainment of sedimentation equilibrium was verified using the program WinMatch. 3 Equilibrium was usually established in 30 -36 h. The equilibrium distributions were then analyzed using Winnl106 4 (19) by performing a global fit of all the data sets to obtain estimates for the association states and equilibrium constants. The program Sednterp, 5 which is based on the method of Laue (20), was used to calculate the protein partial specific volume from the amino acid composition of hATIC-b. This value was calculated to be 0.7304 ml/g at 4°C. Sednterp also calculated the solvent density to be 1.01004 g/ml at 4°C and the monomer molecular mass of ATIC-b to be 67,314. Data in fringe units were converted to molar concentrations using the constant 3.33 fringes⅐mg Ϫ1 ⅐ml Ϫ1 protein and the predicted monomer molecular mass.
Equilibrium Concentration Dependence Assays-Using the determined dissociation constant from the sedimentation equilibrium analytical ultracentrifugation experiments, enzyme stock assay solutions ranging from 8.6 nM to 4.3 M and spanning the percent dimer distribution curve from 6 to 77% were made. Initial stock solutions of hATIC-b were diluted to obtain the final concentration in each assay solution. The concentrations of these initial stocks were determined using Pierce Coomassie ® Plus protein assay reagent with bovine serum albumin as a standard. Each assay solution contained 66 mM Tris-Cl, pH 7.4, 50 mM KCl, and enzyme. The solutions were allowed to reach their oligomeric equilibrium for at least 48 h at 4°C in the cold room. The next day, separate assays for AICAR TFase and IMPCHase activity were performed at 4°C as described above using a 0.1-cm pathlength quartz cuvette. The reaction volume of each assay was 200 l with the final concentrations of each of the substrates, 10-f-FH 4 and AICAR for the AICAR TFase assay and FAICAR for the IMPCHase assay, at 0.5 mM. All assay and substrate solutions were kept at a constant temperature in the cold room before being assayed. Reactions were initiated by the rapid mixing of substrate and assay solution in the cuvette.
Pre ). Tubes were immediately vortexed and kept on ice until all of the samples in the set were collected. Samples were then centrifuged at 4000 ϫ g for 10 min to separate the aqueous and organic phases. The aqueous phase was transferred to a clean tube, lyophilized, and redissolved in 8 l of water for subsequent TLC separation on POLYGRAM ® CEL 300 polyethyleneimine/UV 254 TLC plates (Macherey-Nagel, GmbH & Co. Kg). TLC separation of substrates and products was carried out by slight modification of the method described previously (21).
Pre-steady state burst experiments were fit using the curve-fitting program, KaleidaGraph (Synergy Software), to the burst equation, where A ϭ amplitude of the burst phase, k ϭ observed burst rate, k ss ϭ steady state or linear phase rate, and t ϭ time.

RESULTS
Human ATIC Exists in a Monomer/Dimer Equilibrium-Sedimentation equilibrium analytical ultracentrifugation was used to examine the oligomeric structure of hATIC in solution. A global nonlinear least squares analysis using Winnl106 was carried out on nine data sets obtained from three initial protein concentrations and three rotor speeds. The residuals of the fitting trials suggest that the simplest model that adequately describes the data contains three species with molecular weights corresponding to an ATIC monomer, dimer, and trimer. The formal expression is as follows: where c i is the total fringe displacement at a radial position, r i , c ref is the fringe displacement at a reference position r ref , M is the monomer molecular mass, is the monomer partial specific volume, is the solvent density, is the angular velocity (radians⅐sec Ϫ1 ), R is r i , the universal gas constant, T is the absolute temperature, ϭ r 2 /2, K 1,2 is the apparent monomer/ dimer equilibrium constant (in fringe units) that was common for all data sets, and base is a baseline term for nonsedimentating material.
A typical distribution at 13,300 rpm is shown in Fig. 2. The fit was judged to be good by examination of the residuals (small and random) and minimization of the variance (1.13 ϫ 10 Ϫ2 with 2397 degrees of freedom). From this global fit, the dimer/  monomer dissociation constant was estimated to be 240 Ϯ 50 nM at 4°C using Equation 2. Attempts to globally fit the data using a fitting function containing a common monomer-trimer dissociation constant for all data sets failed to converge as indicated by nonrandomness of the residuals and significant increase in the variance (data not shown). The global analysis thus suggests that the trimer does not reversibly associate with the other species on the time scale of this experiment and is considered thermodynamically heterogeneous.
Effect of Substrates on the Dimerization of ATIC-The obser-  4. Effect of human ATIC concentration on the AICAR TFase activity. A small volume of concentrated substrate solution containing AICAR (500 M) and 10-f-FH 4 (500 M) was combined with a solution containing the given concentration of human ATIC at its oligomeric equilibrium. Upon mixing, the reaction was followed by monitoring the A 298 . The specific activity of the AICAR TFase reaction was plotted against ATIC concentration. Inset, the specific activities were plotted against the percent dimeric ATIC based upon the estimated dissociation constant for ATIC in the presence of AICAR and DDATHF. The dotted line is a linear fit of the data (r ϭ 0.9769). The errors are the S.D. of the specific activities from the raw data.  (4,7,8,11,21). A global nonlinear analysis was carried out as above using two initial concentrations at three rotor speeds for each of the above conditions using Winnl106. The simplest model that fit all the data sets for each condition tested contained three species corresponding to ATIC monomer, dimer, and some heterogeneous trimeric species as before. The results of these studies and their fit statistics are summarized in Table I. As seen in Fig. 3, in the presence of AICAR or 10-f-FH 4 , the percent dimeric species is not significantly shifted from that of apo-hATIC. However, when both AICAR and the folate analog DDATHF are present, there is a considerable shift toward higher percent dimer in the molar species plot showing a relationship between the substrates and the oligomeric state of hATIC. AICAR TFase Is Active as the ATIC Dimer-Knowing that the oligomeric equilibrium is affected by the presence of both AICAR and DDATHF, the concentration dependence of ATIC was investigated to determine whether the amount of dimeric protein has a correlation to the AICAR TFase activity. To study

FIG. 6. Effect of human ATIC concentration on the IMPCHase activity.
A small volume of concentrated substrate solution containing FAICAR (500 M) was combined with a solution containing the given concentration of human ATIC at its oligomeric equilibrium. Upon mixing, the reaction was followed by monitoring the A 248 . The specific activity of the IMP-CHase reaction was plotted against ATIC concentration. Inset, the specific activities were plotted against the percent dimeric ATIC based upon the estimated dissociation constant for ATIC in the presence of AICAR. The dotted line represents a linear fit of the data (r ϭ 0.9952). The errors are the S.D. of the specific activities calculated from the raw data. this, steady state assays were performed within a concentration range of protein based upon the percent dimer curve in Fig.  3. The AICAR and 10-f-FH 4 substrates were concentrated so that the addition of the substrate mixture to the assay solution would not change the total volume by more than 5%. The concentration of substrate used was greater than 100 times the highest enzyme concentration and kept constant throughout so that binding would not be a concern during the experiment. The results shown in Fig. 4 suggest that upon increasing enzyme concentration, the AICAR TFase activity also increases. Correcting the predicted percent dimer for the effect of DDATHF and AICAR on the dissociation constant resulted in a statistically better linear fit of the data (r ϭ 0.9769) (Fig. 4,  inset). Extrapolation of this linear fit to zero percent dimer predicted a specific activity of only 0.03 mol⅐min Ϫ1 ⅐mg Ϫ1 . This suggests that the monomeric form of ATIC is essentially inactive, especially when compared with the dimer's specific activity of 1.1 mol⅐min Ϫ1 ⅐mg Ϫ1 found by extrapolation to 100% dimer.
The result that the dimer is the active form of ATIC for AICAR TFase activity is further supported by the results of pre-steady state burst assays carried out at protein concentrations ranging from 20 to 100 M corresponding to 92 to 97% dimer. Fig. 5 shows a typical pre-steady state AICAR TFase burst experiment using a final concentration of 20 M ATIC carried out at 37°C. A fit of these data using Equation 1 results in an amplitude of 20 M, which suggests that 100% of the AICAR TFase catalytic sites are active at this enzyme concentration.
IMPCHase Is Active in Both the ATIC Monomer and Dimer-An analogous set of experiments for the IMPCHase activity were also carried out on hATIC. The substrate FAICAR was concentrated so that the addition of the substrate mixture to the assay solution would not change the total volume by more than 5%. The concentration of substrate used was greater than 100 times the highest enzyme concentration and kept constant throughout so that binding would not be a concern during the experiment. These results shown in Fig. 6 suggest that upon increasing enzyme concentration, the IMPCHase activity also increases. The percent dimeric ATIC assayed was then calculated using the estimated dissociation constant calculated in the presence of AICAR, because AICAR has been shown to be an inhibitor of IMPCHase activity (22) and may contribute to slightly increased dimerization by binding to ATIC. This correction yielded a slightly statistically better fit (r ϭ 0.9952) than the linear fit using the dissociation constant from hATIC alone (r ϭ 0.9934), but the data points in both cases were within the error limits of each other (data not shown). Extrapolation of the linear fit to 0% dimer indicates a residual specific activity of 0.53 mol⅐min Ϫ1 ⅐mg Ϫ1 suggesting that the monomeric form of ATIC still possesses IMPCHase activity (Fig. 6, inset). This value is about 6-fold less than the predicted specific activity of 3.0 mol⅐min Ϫ1 ⅐mg Ϫ1 for 100% dimeric ATIC. This suggests that the dimeric form is the more active form for the IMPCHase activity.
Determination of the Apparent Dissociation Constant Using Oligomeric Kinetic Data-Another method derived by Kurganov (23) exists to determine the dissociation constant of an enzyme based on kinetic data gathered on a range of protein concentrations. This analysis was carried out on the AICAR TFase kinetic data using the relationship of Kurganov (23) for the case when only the dimeric form of the enzyme is active. This relationship is shown below in Equation 3, where a total is the observed specific activity, a dimer is the specific activity from the dimeric portion of the protein, K 1,2 is the apparent monomer/dimer association constant, and C 0 is the initial enzyme concentration (23). Graphing the data in the coordinates (a total ) Ϫ1 versus (C 0 a total ) Ϫ1/2 produced a linear trend with a linear correlation coefficient of 0.9950 for the linear regression fit (data not shown). Using the slope and intercept values from the fit, the values of a dimer and K 1,2 were calculated to be 1.06 mol/min/mg and 0.035 nM Ϫ1 , respectively. This value of K 1,2 gives an apparent dissociation constant of 28 nM, which is in agreement with the dissociation constant calculated from the centrifugation data of hATIC in the presence of AICAR and DDATHF. Attempts to analyze the IMPCHase kinetic data were unsuccessful because of the difficulty in gathering data over a large enough concentration range required to satisfy the conditions where both the monomer and dimer possess enzymatic activity (23). DISCUSSION Our interest in understanding the structure-function relationships for ATIC has led us to characterize the oligomeric state of human ATIC. Using sedimentation equilibrium analytical ultracentrifugation we have shown that human ATIC exists in a monomer/dimer equilibrium with an apparent dissociation constant of 240 Ϯ 50 nM at 4°C. The two enzyme cross-linking studies reported previously (10,11) support this finding, because both groups observed dimeric ATIC and neither observed any higher order species.
The effect of substrates proved interesting in that neither AICAR nor 10-f-FH 4 alone significantly changed the apparent dissociation constant, but when AICAR was present along with DDATHF, a folate analog of 10-f-FH 4 , the apparent dissociation constant was reduced 10-fold, resulting in tighter dimer formation. Thus it appears that although neither of the substrates alone provides further stabilization of the dimer, when present together, there is considerable additional stabilization. This suggests that these two substrates bind ATIC in such a manner as to either directly bridge the dimeric interface of the homodimer or to induce a conformational change to increase the interactions between the monomer subunits.
Recent crystallographic studies on both the avian and human forms of ATIC by the Wilson group at The Scripps Research Institute suggest that the former is probably the case. Their findings have shown that ATIC crystallizes as a homodimer and that the presumed substrate binding site for the AICAR TFase activity consists of residues from both monomeric units of ATIC. The IMPCHase substrate binding site is completely localized in one monomeric unit, away from the dimeric interface. 6 This led us to investigate whether there is a fundamental relationship between dimerization and catalytic activity for the AICAR TFase and IMPCHase reactions. As seen in Fig. 4, there is an enzyme concentration dependence on AICAR TFase activity of ATIC, such that extrapolation of the specific activity to 0% dimer predicts practically no residual AICAR TFase activity. The values for the percent dimer were calculated using the estimated dissociation constant found with both folate and AICAR present, because both substrates are required for catalysis. Thus it appears that the dimeric form of ATIC is strongly preferred for AICAR TFase activity. This finding correlates with the structural data mentioned above in that the binding sites for the AICAR and folate substrate appear to be at the dimer interface and made up of constituents from both monomer subunits. Also, the pre-steady state burst experiments on the AICAR TFase activity, which were carried out at concentrations where ATIC is mostly dimeric, showed 100% of the catalytic sites were active. If the dimeric form of ATIC were not an active form of the enzyme, this number would be expected to be much lower.
As a check of our results, we analyzed our AICAR TFase kinetic data using a method derived by Kurganov (23). This was important to show that the assumption of using the estimated dissociation constant as determined in the presence of both AICAR and DDATHF was justified when plotting the AICAR TFase kinetic data, as well as to substantiate the value of the monomer/dimer dissociation constant obtained from the sedimentation equilibrium studies. This was of particular significance, because the value of 31 Ϯ 9 nM calculated from the analytical ultracentrifuge data was near the limit of this technique. Using the Kurganov method, the estimated dissociation constant was calculated to be 28 nM, well within the error calculated from the sedimentation equilibrium method for the condition when both AICAR and DDATHF were present.
An enzyme concentration dependence was also found for the IMPCHase specific activity (Fig. 6); in this case, however, both the monomer and dimeric forms of ATIC appear to possess activity, with the dimeric form being more active. This result also fits the structural data, because the IMPCHase substrate binding site is completely localized in the monomeric subunit as stated above. The fact that the IMPCHase activity was greater when ATIC was in the dimeric form may possibly be attributed to the dimeric form of ATIC keeping the IMPCHase active site in a more active conformation. Perhaps the IMP-CHase activity is more active in the dimeric form than the monomeric form to help facilitate the conversion of FAICAR formed from the AICAR TFase reaction, because it is the dimeric form that is required to provide the FAICAR for catalysis to IMP at the IMPCHase site.
The linking of the AICAR TFase and IMPCHase activities on the same polypeptide raises the question of whether these two activities are kinetically coupled to one another through either substrate channeling or communication between the two catalytic domains. Szabados and Christopherson (24) observed a decrease in the formation of [ 3 H]IMP from [ 3 H]AICAR when exogenous unlabeled FAICAR was added to the bifunctional enzyme reaction and interpreted this as indicating that substrate channeling does not occur in ATIC (24). However, those experiments were not in themselves enough to completely justify that conclusion. A detailed transient-phase kinetic study of the AICAR TFase and IMPCHase activities of ATIC is currently underway in our laboratory to unequivocally answer the substrate channeling question. Communication between the two active sites also seems unlikely, because the construction of truncation mutants of ATIC resulted in two separate, fully active, functional domains containing AICAR TFase activity on the C-terminal end and IMPCHase activity on the N-terminal end (8). This is further supported by the fact that IMPCHase inhibitors have no effect on the AICAR TFase kinetics (25). Regardless of whether channeling occurs in ATIC, there is yet an advantage to having these two activities located on the same polypeptide. Recent studies by ourselves 7 and others (25) have independently shown that the AICAR TFase reaction favors the reverse direction. The tethering of these two activities puts the IMPCHase and the AICAR TFase domains in close proximity to one another. The decreased distance between these two activities should help draw the overall reaction in the forward reaction by allowing the AICAR TFase domain to provide an increased local concentration of the intermediate FAICAR to the IMPCHase domain for conversion to IMP. If these two activities were located on separate proteins, the efficiency of the system would be reduced because of the dilution effects of diffusion through solvent and also the possible conversion of FAICAR back to AICAR. Because it is the dimeric form of ATIC providing the FAICAR for the IMPCHase domain, a more active IMPCHase activity in the dimeric form would help to make the overall reaction more efficient.
In conclusion, these studies of human ATIC provide an insight into the structure-function relationships of this important bifunctional enzyme in the de novo purine biosynthetic pathway. The results presented here show that ATIC exists in a monomer/dimer equilibrium but that the dimeric form is required for AICAR TFase activity whereas the IMPCHase possesses catalytic activity in both the monomeric and dimeric forms with the dimeric form being more active. These findings will be of greatest importance in postulating a catalytic mechanism for both activities of ATIC, especially for the AICAR TFase reaction. Also, these results could possibly be used in the design of inhibitors that do not target the active sites of this protein but those that inhibit dimerization of ATIC and thus inhibit the AICAR TFase reaction.