The Two Toxoplasma gondii Hypoxanthine-Guanine Phosphoribosyltransferase Isozymes Form Heterotetramers*

Two isozymes of the purine salvage enzyme hypoxan-thine-guanine phosphoribosyltransferase (HGPRT) of the apicomplexan protozoan Toxoplasma gondii are encoded by the single HGPRT gene as a result of differential splicing. Western blotting of total T. gondii protein shows that both isozymes I and II, which differ by 49 amino acids, are expressed. Both form enzymatically active homotetramers when overexpressed in Escherichia coli . The specific activity of HGPRT-I is five times that of HGPRT-II. When both isozymes are co-expressed in E. coli , HGPRT-I z HGPRT-II heterotetramers form. The predominant heterotetramer has enzymatic activity similar to HGPRT-II, and gel filtration chromatography demonstrates that its size is intermediate between the sizes of HGPRT-I and HGPRT-II. Mass spectrometric analysis of cross-linked homo- and heterotetramers re-veals species of distinct molecular mass for HGPRT-I, HGPRT-II, and HGPRT-I z HGPRT-II and suggests that the predominant heterotetramer consists of one HG-PRT-I subunit and three HGPRT-II subunits. The impli-cations of this finding are discussed. BIO 101, Inc.) and then ligated with T4 DNA ligase (16 h, 15 °C). Competent E. coli XL1 Blue cells were transformed to kanamycin resistance with the ligated DNAs and grown on 2x-YT (16 g/liter tryptone, 10 g/liter yeast extract, and 5 g/liter NaCl) agar plates containing 25 m g/ml kanamycin. Colonies were screened for the presence of HGPRT-I and HGPRT-II inserts by restriction enzyme ( Nde I and Bam HI) analysis of plasmid DNA. Two positive clones were named pET9a-C1 and pET9a*

Encephalitis caused by the apicomplexan protozoan Toxoplasma gondii is the second leading cause of death among patients with AIDS (1). Much effort has been expended on the characterization of this parasite and on the discovery and development of new drugs to treat T. gondii infections (2,3). A widely recognized drug target in parasitic protozoa is the enzyme hypoxanthine-guanine phosphoribosyltransferase (HG-PRT 1 ; EC 2.4.2.8) (4,5). HGPRT catalyzes the Mg 2ϩ -dependent conversion of hypoxanthine, guanine, or xanthine and ␣-D-5phosphoribosyl-1-pyrophosphate (PRPP) to purine nucleotides and inorganic pyrophosphate. It is a key purine salvage enzyme in T. gondii, which cannot carry out the de novo synthesis of purine nucleotides required for growth and replication (6,7).
The cloning of T. gondii HGPRT revealed the presence of two cDNAs, differing by 147 nucleotides, that appear to result from differential splicing of the nascent transcript from the single HGPRT gene (8,9). It is the smaller 230-amino acid isozyme I (HGPRT-I), which is homologous to human HGPRT, that we (11,12) 2 and others (13) have crystallized. Enzymatically active HGPRT-I is a homotetramer (12). The larger T. gondii HGPRT isozyme, HGPRT-II, has a 49-amino acid insertion following Glu 7 (GenBank TM accession number U10083). How the three-dimensional structure of HGPRT-I is altered to accommodate the insertion in HGPRT-II, which is located at a subunit interface within the HGPRT tetramer, is not known.
T. gondii HGPRT is the only HGPRT known that may exist as two isozymes. From a drug design perspective, therefore, it is important to know whether Toxoplasma expresses HG-PRT-II as a stable enzyme, and if so whether HGPRT-I or HGPRT-II (or both) is the true drug target in Toxoplasma. If both HGPRT-I and HGPRT-II are expressed in Toxoplasma, do they form homotetramers only, or are they capable of forming heterotetramers as well? If so, do they form a preferred heterotetramer or a statistical mixture of all possible heterotetramers? How do the enzymatic activity and sensitivity to inhibitors of HGPRT heterotetramers compare with HGPRT-I and HGPRT-II homotetramers ? We report here our initial investigations into these questions, which show that T. gondii does express stable HGPRT-II, that the HGPRT-II homotetramer possesses enzymatic activity distinct from that of HGPRT-I, that HGPRT-I and HGPRT-II form a particular heterotetramer when both are co-expressed in Escherichia coli, and that this heterotetramer possesses enzymatic activity similar to that of the HGPRT-II homotetramer.

HGPRT Expression
Subcloning into pET9a-Plasmids pET15b-C1 (previously pETC1; Ref. 8) and pET15b-C3 (pETC3; Ref. 8), which contain the T. gondii HGPRT isozymes I and II coding sequences, respectively, inserted into vector pET15b (Novagen, Inc.; Ref. 14), were digested with NdeI and BamHI overnight at 37°C. pET9a was digested as well. pET9a is similar to pET15b but lacks the N-terminal His 6 tag and contains a kanamycin rather than an ampicillin resistance marker. Digested DNAs were separated by electrophoresis on a 1% agarose gel; the appropriate fragments were extracted from the gel (GeneClean, BIO 101, Inc.) and then ligated with T4 DNA ligase (16 h, 15°C). Competent E. coli XL1 Blue cells were transformed to kanamycin resistance with the ligated DNAs and grown on 2x-YT (16 g/liter tryptone, 10 g/liter yeast extract, and 5 g/liter NaCl) agar plates containing 25 g/ml kanamycin. Colonies were screened for the presence of HGPRT-I and HGPRT-II inserts by restriction enzyme (NdeI and BamHI) analysis of plasmid DNA. Two positive clones were named pET9a-C1 and pET9a-* This work was supported in part by National Institutes of Health Grants AI39952 (to D. W. B.) and AI30279 (to James R. Piper). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: HGPRT, hypoxanthine-guanine phosphoribosyltransferase; HGPRT-I, T. gondii HGPRT, isozyme I (short form); HGPRT-II, T. gondii HGPRT, isozyme II (long form); MALDI-TOF, matrix-assisted laser desorption ionization-time of flight; PRPP, ␣-D-5-phosphoribosyl 1-pyrophosphate; PAGE, polyacrylamide gel electrophoresis.
C3. Open reading frames in each plasmid construct were confirmed by automated DNA sequencing (PRISM Dye Terminator kit/PRISM 377 sequencer, Applied Biosystems, Inc.) to ensure that no sequence artifacts were introduced during the subcloning process.

HGPRT Purification and Characterization
Purification and Enzymatic Assay-Isozyme II of T. gondii HGPRT was purified from the pET15b expression culture by Ni 2ϩ -agarose affinity chromatography followed by thrombin digestion (to remove the N-terminal His 6 tag) and gel filtration, essentially as described previously for HGPRT-I (12). HGPRT-II was found to be unusually susceptible to thrombin digestion. Consequently, proteolysis was carried out with 5 units of thrombin/ml for 2 h (rather than 4 units/ml overnight) on ice and was followed by passage through a small benzamidine-agarose affinity column to remove the thrombin. Additional protease inhibitors were added to the purified enzyme. T. gondii HGPRT heterotetramers were purified from the co-expression cultures in the same manner. HGPRT kinetics were measured spectrophotometrically at 37°C as described (12), in a buffer containing the purine base, PRPP, 100 mM Tris⅐HCl, pH 8.0, 20 mM MgCl 2 , 0.1 mM EDTA, and 0.1 mg/ml bovine serum albumin. V max values were converted to k cat values using protein concentrations determined by Bradford assay (bovine ␥ globulin standard; Ref. 15).
Subunit Ratio and Cross-linking-Discontinuous SDS-polyacrylamide gel electrophoresis (PAGE), performed according to Laemmli (16), was used to assess HGPRT purity, to determine approximate subunit molecular masses, and to assess the extent of cross-linking reactions. HGPRT-I to HGPRT-II ratios were determined by densitometry (Shimadzu CS-9000 dual wavelength flying spot scanner, single lane scans at 595 nm) of gels stained with Coomassie Brilliant Blue G-250 (Pierce Gelcode Blue). Cross-linking reactions were performed by incubating 100 l of HGPRT (2.5 mg/ml) in phosphate-buffered saline and 2 l of disuccinimidyl suberate (20 mM in Me 2 SO, Pierce) for 30 min at room temperature. Reactions were quenched by addition of 1 l of 1 M ethanolamine.
Western Blotting of T. gondii Total Soluble Protein-Purified recombinant T. gondii HGPRT-I was separated on 12% SDS-PAGE, stained with 0.1% Coomassie Brilliant Blue G-250, destained, and rinsed extensively in H 2 O. The band at ϳ27 kDa was cut from the gel. Polyclonal rabbit antisera were prepared by HRP Inc. (Denver, PA). The minced gel containing ϳ125 g of protein was injected subcutaneously into each of two rabbits along with Freund's adjuvant. Two additional injections were made, and the rabbits were bled for serum production. Preimmunization bleeds were also taken. T. gondii parasites (RH strain tachyzoites) were lysed by freeze-thawing and brief sonication in 25 mM Tris⅐HCl, pH 7.5, 10 mM MgCl 2 , 1 mM PRPP, 1% Nonidet P-40. The lysate was centrifuged (10,000g, 4°C, 10 min). 12 g of T. gondii soluble protein was separated by 12% SDS-PAGE; recombinant T. gondii HG-PRT-I and HGPRT-II (400 ng each) and molecular mass markers were included on the gel as standards. The gel was transferred to Hybond-ECL (Amersham Pharmacia Biotech) in Towbins buffer (25 mM Tris⅐HCl, 192 mM glycine, 20% (v/v) methanol, pH 8.3, at 100 V for 1 h). A 1:1000 dilution of the rabbit polyclonal anti-HGPRT-I antiserum was used as the primary antibody, and a 1:1500 dilution of the Enhanced Chemiluminesence Western Kit (Amersham Pharmacia Biotech) antirabbit IgG antibody was used as the secondary antibody. Preimmunization serum was used in control experiments.
Mass Spectrometry and Amino Acid Sequencing-MALDI-TOF mass spectra were obtained on a Voyager Elite mass spectrometer (positive mode) with delayed extraction technology (PerSeptive Biosystems). The acceleration voltage was set at 25 kV, and 10 -50 laser shots were summed. The matrix was sinapinic acid (Aldrich) dissolved in CH 3 CN/ 0.1% CF 3 CO 2 H (1:1). The spectrometer was calibrated with apomyoglobin or bovine serum albumin. Samples were diluted 1:10 with matrix before pipetting 1 l onto a smooth plate. N-terminal sequencing was done by automated Edman degradation on a gas-phase microsequencing system (model PI 2090E, Beckman). The amino acid residue released in a given cycle was identified from the difference chromatogram (comparison with the previous cycle).

Expression, Purification, and Characterization of HGPRT-II-
We showed previously that expression of T. gondii HG-PRT-I in E. coli (pET15b vector) provides large quantities of homogeneous, enzymatically active protein (12). HGPRT-II was expressed using the same approach. It was purified to apparent homogeneity by Ni 2ϩ -agarose affinity chromatography, proteolytic removal of the N-terminal His 6 tag, and gel filtration chromatography. Isozyme II had enzymatic properties similar to but distinct from isozyme I. As shown in Table I, HGPRT-II had about one-fifth the specific activity of HGPRT-I regardless of whether hypoxanthine, guanine or xanthine was the substrate. The Michaelis constants of the two isozymes were similar, with the PRPP K m of HGPRT-II tending to be higher than that of HGPRT-I. Also, the guanine K m of HG-PRT-II was six times higher than that of HGPRT-I, with the result that the catalytic efficiency of the isozymes differed by 30-fold with this substrate. These kinetic differences between HGPRT-I and HGPRT-II agree in broad terms (except for guanine) with those reported previously (9).
The subunit molecular mass of HGPRT-II, as determined by SDS-PAGE, was 33,500 Da. Both it and HGPRT-I (27,500 Da) migrated at a somewhat higher position than that predicted from their calculated molecular masses. Nonetheless, mass spectrometric analysis (MALDI-TOF) of both isozymes confirmed that they possessed the correct mass (for HGPRT-I, the observed mass was 26,668 Da and the calculated mass was 26,667 Da; for HGPRT-II, the observed mass was 31,767 Da and the calculated mass was 31,766 Da; masses include an N-terminal peptide, Gly-Ser-His, that remains after thrombolytic removal of the His 6 tag).
HGPRT-II eluted much earlier than HGPRT-I from a gel filtration column, well in excess of that expected from molecular mass differences, suggesting that it possessed a higher oligomeric state than HGPRT-I (Fig. 1). Comparison of its elution volume with those of standards of known size revealed that HGPRT-II had a Stokes radius of 49.2 Å, compared with 38.2 Å for HGPRT-I (17). We had determined previously from its Stokes radius that HGPRT-I was a tetramer (Perrin shape factor F, 1.04; axial ratio, ϳ1.9:1; assumes 0.35 g of H 2 O/g of HGPRT-I), a result that has been verified repeatedly by the observation of this aggregation state exclusively in more than eight distinct T. gondii HGPRT crystal structures (axial ratio ϳ1.6:1; Refs. 11-13). 2,3 When the shape factor was calculated for possible HGPRT-II aggregation states (dimer through octamer), only an octamer appeared physically reasonable (F, 1.03; axial ratio, ϳ1.7:1) (18). Other possible aggregation states afforded axial ratios of Ͼ3.5:1. These results suggested that HGPRT-II either adopts a dramatically more open conformation than HGPRT-I, or more likely that it, like HGPRT-I, forms a tetramer, but one prone to further aggregation to an octamer under the chromatography conditions.
Both HGPRT-I and HGPRT-II Are Expressed in Toxoplasma-Immunological detection was used to determine whether both T. gondii HGPRT cDNAs give rise to stable protein in the parasite. A polyclonal rabbit antiserum was raised against homogeneous recombinant HGPRT-I. Western blotting showed that the antiserum cross-reacted well with recombinant HG-PRT-II, as expected from the high similarity in amino acid sequence. A Western blot of total T. gondii soluble protein using this antiserum for detection is shown in Fig. 2. It was clear that approximately equal amounts of both HGPRT-I and HGPRT-II subunits were present in Toxoplasma. Control blots using preimmunization serum demonstrated that the bands in Fig. 2 arose specifically from HGPRT. The possibility that the band we attributed to HGPRT-I in the parasite actually arose from contaminating human HGPRT, an inevitable low level contaminant given that the parasites were grown in human fibroblasts, was excluded because a control blot showed that the antiserum did not cross-react with recombinant human HGPRT.
HGPRT-I and HGPRT-II Form Heterotetramers When Coexpressed in E. coli-We took advantage of the differences between the pET15b and pET9a vectors (N-terminal His 6 tag and ampicillin resistance versus no tag and kanamycin resistance) to express both T. gondii HGPRT isozymes (one tagged and one untagged) simultaneously in E. coli. The only way the untagged HGPRT subunit could bind to a Ni 2ϩ -agarose affinity chromatography column would be as part of a heterotetramer with the tagged subunit.
Maintenance of the double antibiotic selection allowed tagged HGPRT-I (pET15b) and untagged HGPRT-II (pET9a) to be co-expressed in E. coli BL21(DE3). The reciprocal experiment (untagged HGPRT-I and tagged HGPRT-II) was also successful. The soluble fraction of the cell lysate was passed over Ni 2ϩ -agarose, and nonspecifically bound proteins were eluted with a 150 mM imidazole wash. Bound protein was then eluted with 450 mM imidazole, which was analyzed by SDS-PAGE and densitometry. Regardless of whether HGPRT-I or HGPRT-II carried the His 6 tag, specifically bound protein consisted of a 1:3 ratio of the HGPRT-I and HGPRT-II subunits (Fig. 3), even when the relative expression levels of HGPRT-I and HGPRT-II differed. This is the first evidence that HGPRT heterotetramers can and do form when both isozymes are heterologously expressed in E. coli. Control experiments showed that untagged HGPRT-I, HGPRT-II, or HGPRT-I⅐HGPRT-II heterotetramers, by themselves, do not bind to Ni 2ϩ -agarose.
When HGPRT-II bore the His 6 tag, both proteins were expressed at moderate, approximately equal levels (Fig. 3, top  panel). The majority of the HGPRT bound specifically to Ni 2ϩagarose as the 1:3 HGPRT-I⅐HGPRT-II heterotetramer. Excess untagged HGPRT-I, which did not bind, was enzymatically active. The purified heterotetramer had specific enzymatic activity much more similar to HGPRT-II than HGPRT-I (Table I).
In the reciprocal experiment, tagged HGPRT-I was barely detectable in cell lysates 3 h after isopropyl-1-thio-␤-D-galactopyranoside induction, whereas untagged HGPRT-II was expressed at very high levels (Fig. 3, bottom panel). As before, the 1:3 HGPRT-I⅐HGPRT-II heterotetramer bound specifically to Ni 2ϩ -agarose, whereas the large excess of untagged HGPRT-II did not. Unbound co-expressed HGPRT-II appeared to have a lower molecular mass by SDS-PAGE (31,500 Da) compared with HGPRT-II expressed alone (33,500 Da, after thrombin cleavage of the His 6 tag). Purification of unbound HGPRT-II by gel filtration chromatography (Stokes radius, 50.2 Å) provided a sample of Ͼ95% purity that possessed enzymatic activity equivalent to that of HGPRT-II expressed alone. N-terminal sequencing (ASKPIEESR) and MALDI-TOF mass spectrometry (observed mass was 31,408 Da) suggested that it comprised amino acids 2-279 of HGPRT-II (calculated mass was 31,353 a HGPRT-II-tagged. The Ni 2ϩ -agarose affinity-purified heterotetramer was further purified by gel filtration chromatography (Fig. 1, bottom  panel).
b Based on comparative specific activity measurements of HGPRT-II and the HGPRT-I⅐HGPRT-II heterotetramer. PRT-I with the tag remnant migrate at slightly higher than expected masses. SDS-PAGE yields at best only approximate molecular masses, because of its sensitivity to a variety of factors. Nevertheless, the mass spectral results for all three proteins agree very well with the masses calculated from their amino acid compositions. Additional Evidence That HGPRT-I⅐HGPRT-II Heterotetramers Form-Although the results of the co-expression experiments presented above, by their very design, provided strong evidence that HGPRT heterotetramers were formed, we sought additional confirmatory evidence. As shown in Fig. 1, the Ni 2ϩagarose-bound HGPRT heterotetramer had a gel filtration elution volume more similar to the elution volume of HGPRT-II than HGPRT-I. A control experiment with a physical mixture of the two homotetramers ( Fig. 1) showed conclusively that the peak of intermediate mobility was due to a distinct heterotetrameric species (Stokes radius, 47.7 Å).

TOF mass spectrometry. As shown in
The mass spectrum of cross-linked HGPRT-I⅐HGPRT-II (HG-PRT-I untagged, observed subunit mass, 26,218 Da; calculated mass (residues 2-230), 26,255 Da; HGPRT-II tagged and thrombin-cleaved, subunit mass, 31,667 Da) is notable for the absence of cross-linked HGPRT-I peaks (Fig. 4, stars) and the presence of two novel peaks (Fig. 4, A and D). Possible expected and observed masses for a cross-linked heterotetramer are summarized in Table II. No peak was observed for the HG-PRT-I homodimer (Fig. 4, left star). By contrast, a new peak for the HGPRT-I/HGPRT-II heterodimer was observed (peak A). Peaks for the HGPRT-II/HGPRT-II homodimer and the HG-PRT-II⅐HGPRT-IIЈ heterodimer were also observed. Similarly, of the possible cross-linked trimers, only those corresponding to the HGPRT-II homotrimer and the HGPRT-II/HGPRT-II/HG-PRT-IIЈ and HGPRT-II/HGPRT-II/HGPRT-I (peak D) heterotrimers were observed. The HGPRT-I homotrimer was not observed (Fig. 4, right star). The simplest interpretation of these results is that co-expressed HGPRT consists of a tetramer that contains no more than one HGPRT-I subunit, i.e. a 1:3 HGPRT-I⅐HGPRT-II heterotetramer. DISCUSSION HGPRT has been characterized from many mammalian, protozoal, and bacterial organisms. To the best of our knowledge, only the HGPRT from T. gondii has been found to exist as two isozymes. This finding is especially interesting for two reasons. First, T. gondii HGPRT is a novel drug target for the treatment of opportunistic toxoplasmosis in AIDS patients (4,5). Because there are two HGPRTs in T. gondii, both should be evaluated as drug targets. Second, the two isozymes, HGPRT-I and HG-PRT-II, are encoded by a single HGPRT gene (9). Two research groups each isolated from T. gondii cDNA libraries two mature HGPRT transcripts that appear to result from differential  Table II. Note that the different proteins ionize with different efficiencies. In particular, HGPRT-I appears to ionize much more efficiently than HGPRT-II, and the HGPRT-II-containing samples consist predominantly of full-length HGPRT-II, with a minor amount (Ͻ17%) of HGPRT-IIЈ. The m/z ratio of the peaks are indicated. In the spectrum of the HGPRT-I⅐HGPRT-II heterotetramer, six peaks (A-F) listed in Table II are indicated, and the positions of unobserved peaks attributable to HGPRT-I cross-linked to itself are marked by stars.  Fig. 4. splicing of the nascent transcript (8,9). We thought at first that the isolation of the HGPRT-II cDNA was a cloning artifact, perhaps as a result of inefficient or incomplete processing of the nascent HGPRT-II mRNA to the mature, HGPRT-I mRNA in T. gondii. This thought was reinforced by the fact that it is the smaller HGPRT-I that is homologous to all other known HG-PRTs, including its closest known relative, Plasmodium falciparum HGPRT (19,20), and that all other organisms possess only one HGPRT protein.
Our experiments with the HGPRT-II isozyme show that it possesses high enzymatic activity when expressed in E. coli, like HGPRT-I. Donald et al. (9) previously reported enzymatic activities for recombinant HGPRT-I and HGPRT-II that agree relatively well with our values (Table I). Because of purification difficulties with HGPRT-II, however, these workers were not able to state conclusively whether the differences they observed between HGPRT-I and HGPRT-II were real. Our straightforward purification of HGPRT-II now shows that it is indeed enzymatically different from HGPRT-I.
Western blotting of T. gondii total soluble protein, using polyclonal antiserum raised against recombinant HGPRT-I, demonstrated that both isozymes are expressed as protein in the parasite. Our findings indicate, therefore, that the isolation of two HGPRT cDNAs from T. gondii was not an artifact but rather that this organism actually has two functional HGPRTs. Although it is not clear why Toxoplasma has two HGPRTs, the significance for drug design is clear: recombinant HGPRT-I and HGPRT-II have different catalytic rate and Michaelis constants (especially for guanine and PRPP), and thus they may have different susceptibility to inhibitors. Also, certain mutations in human HGPRT bestow cooperativity upon this usually noncooperative enzyme (10,21). Experiments are in progress to determine whether HGPRT-II or the heterotetramer exhibits cooperativity.
Because T. gondii HGPRT exists as two isozymes, each of which is a homotetramer when expressed in E. coli, a natural question to ask is whether the isozymes are capable of forming heterotetramers? Our co-expression experiments have provided an affirmative answer to this question. Indeed, when HGPRT-I and HGPRT-II are simultaneously expressed in E. coli, a heterotetramer comprising one HGPRT-I subunit and three HGPRT-II subunits forms.
The evidence for this unexpected 1:3 heterotetramer is as follows. First, the co-expression experiment was designed to provide heterotetramers in which only one isozyme carried a purification tag. Affinity-purified heterotetramers were separated by SDS-PAGE, and densitometry showed that the HG-PRT-I and HGPRT-II subunits were present in a 1:3 ratio, regardless of whether HGPRT-I or HGPRT-II was tagged and regardless of the very different relative and absolute expression levels of the two isozymes (Fig. 3). Second, gel filtration chromatography of the affinity-purified heterotetramers revealed predominantly a single species, with a mobility intermediate between those of HGPRT-I and HGPRT-II but closer to HGPRT-II (Fig. 1). Third, when the heterotetramers were cross-linked and analyzed by MALDI-TOF mass spectrometry, cross-linked HGPRT-I subunits were not observed, but crosslinked HGPRT-I and HGPRT-II subunits were observed ( Fig. 4 and Table II).
In E. coli, formation of the 1:3 HGPRT-I⅐HGPRT-II heterotetramer is accompanied by formation of the homotetra-mers. It is intriguing that the 1:3 HGPRT-I⅐HGPRT-II heterotetramer has enzymatic activity and physical properties similar to HGPRT-II (Table I and Fig. 1). It is not clear why this particular heterotetramer is the one formed in E. coli, rather than a statisical mixture of heterotetramers. It is unknown whether HGPRT heterotetramers form in T. gondii.
Our gel filtration results suggest that the HGPRT-II homotetramer and the heterotetramer may form larger oligomers. It is unclear at this point whether these enzymes are truly octamers, as their Stokes radii would suggest, or whether they are just tetramers prone to further aggregation. Certainly, the fact that the HGPRT-I dimer is not observed in cross-linked samples of the heterotetramer suggests that the latter interpretation is correct. Octameric HGPRT would also be unprecedented. Analytical ultracentrifugation and crystallization experiments with HGPRT-II and the heterotetramer are planned to address this issue. Finally, further studies are contemplated to express tagged HGPRT-I and HGPRT-II in T. gondii itself, which will allow for the confirmation that heterotetramers form in situ. If so, their purification and characterization will be undertaken.