Chinese Hamster Hypoxanthine-Guanine Phosphoribosyltransferase PURIFICATION, STRUCTURAL, AND CATALYTIC PROPERTIES*

SUMMARY Hypoxanthine-guanine phosphoribosyltransferase from Chinese hamster brain, liver, and V79 tissue culture cells appears to have identical structural and catalytic properties. The enzyme has been purified 540-fold to apparent homogeneity from Chinese hamster brain. The native molecular weight is 78,000 to 85,000 determined by Sephadex G-100 column chromatography and acrylamide gel electrophoresis. The enzyme appears to consist of three subunits of molecular weight 25,000 determined by sodium dodecyl sulfate acrylamide gel electrophoresis. Electrofocusing and acrylamide gel electrophoresis demonstrate the presence of at least three isozymes. The enzyme is remarkably stable at 85” if first incubated in 1 mM 5-phosphoribosyl 1-pyrophosphate. The enzyme is active from pH 5.5 to 11 with maximum activity at pH 10. The enzyme displays Michaelis-Menten kinetics with apparent Michaelis constants for hypoxanthine, guanine, and phosphoribosylpyrophosphate of 0.52, 1.1, and 5.3 pM, respectively. multistep

I'urine nucleotides are formed either by a multistep de no00 synthetic pathway, or by salvage enzymrs which enable the utilization of preformed purinc bases. IIypoxarlthirlt-guarlilic phosphoribosyltransferasc ( # To whom requests for reprints should be addressed. 1 The abbreviation used is: PRPP, 5.phosphoribosyl l-pyrophosphate. pathways alone, whole ailimals apparently require both. Even in Lesch-Nyhan syndrome patirnts who show marked deficiency of hypoxanthine-guaninc phosphoribosyltransferase, low erlzyme activity remains (I-3).
Total deletion of this salvage enzyme would probably be lethal since some cells are incapable of de 1~0~0 purine synthesis (4-6).
The dependence upon salvage pathways of leukocytes forms the rationale for chemotherapy in the treatment of leukemia patients (7). Purine base analogs are administered t,o kill prcfcrent'ially actively dividing cells that have an absolute rcquirement for preformed purincs.
Treatmeilt is often initially successful, but after a time the leukemic cells often appear resistant to the drug(s).
It would be valuable to know whcthcr altered forms of llypo~alltlliric-guallirle phosphoribosyltrarlsferase could account for this resistance.
Also, a better understanding of the mechanisms of purinc regulation rnight suggest improved methods of cancer chemotlic~rapy.
Since it is difficult to study human patients, we have chosen to investigate mechanisms of drug resistance and purinc regulation in tissue culture cells. Chinese hamster cells offer numerous advantages over human cells for genetic and regulatory studies (8). XI&hods for selecting drug-resistant variants are well documented, and we have produced phenotypically different variants in many different ways. Hyposanthineguanine phosphoribosyltransferase activity in drug-resistant variants ranges from total absence to normal levels. An analysis of thcsc variants requires knowledge of the properties and characteristics of the enzyme. In this paper, WC present the purification procedure for a homogeneous enzyme from Chinese hamster braill and a comparison of the physical and biochemical properties of the brain enzyme with the enzyme prepared from liver and tissue culture cells. in an Amicon ultrafiltration cell with a PM-10 membrane.

DEAE-Sephadex
Fractiolz-This was the same procedure as that used for the brain enzyme except that the column was larger (2.5 X 25 cm).
After application of the enzyme, the column was washed with 85 ml of enzyme buffer, and the enzyme was eluted with a 400-ml KC1 gradient.
The active fractions were pooled and concentrated but no PRPP was added. pH 4.5 Supernalarlt-The procedure was the same as that for the brain enzyme, except thai a 0.05 volume of 1 M sodium acetate and a 0.05 volume of 2 M Tris-HCl were added. 85" Supernatant-The procedure was the same as that for the brain enzyme. The gel and buffer system of Laemmli (12) was used with minor modifications.
A separating gel of 12.5% acrylamide and a stacking gel of 4.5c/;, acrylamide were used. The gels were run at room temperature at, 100 volts until the dye band entered the separating gel, and at 200 volts until the dye band reached the bottom of the gel.

No?rde/Laturing Gel Electrophoresis
A modification of the procedure of Davis (13) was used to prepare separat,ing gels of 4 to 9% acrylamide and st,acking gels oi3yo acrvlamide.
The electrode buffer CDH 8.5) contained 10 rnM Trisbase, 75 mM glycine, and 2 mM 2-mercaptoethanol. Gels were run at 5-10" at 200 volts until the dye band-entered the separating gel, then at, 400 volts until the dye band reached the bottom of the gel. The sections of the gel to be assayed were cut into 2.5.mm slices, placed into tubes, and 100 ~1 of the standard assay mixture were added.
These were then incubated at 37" for 15 min. and the products were separated as described above  hours. IMP synthesis catalyzed by purified enzyme (brain 85" supernatant) is linear for 45 min, and linearity is extended to 90 min if the protein concentration in the assay is increased by the addition of 0.1 mg per ml of bovine serum albumin.
More than 90% of the available substrate can be used without affecting the rate of the reaction.
The rate of IMP formation is also directly proportional to protein concentration at all steps of the enzyme purification.
The specific activity of enzyme fractions is determined from the slope of a plot of the rate of IMP formation versus protein concentration. Table I summarizes the purification of hypoxanthine-guanine phosphoribosyltransferase from 24 Chinese hamster brains. The first two steps, low pH and 65" heat, eliminate 90% of the protein but yield only a 3.2.fold purification.
This value is artificially low since cumulative recovery almost doubles after the next step, ammonium sulfate fractionation.
The enzyme is adsorbed to DEAE-Sephadex in low ionic strength enzyme buffer and eluted by a linear salt gradient in an irregular peak from 30 to 70 mM KC1 as shown in Fig. 1. No activity is removed during the initial wash, nor can additional activity be removed by KC1 concentrations up to 0.7 M. Generally, ad-Sodium dodecyl sulfate acrylamide gel electrophoresis demonstrates that incubation of the DEAE-fraction for 10 min at 85" produces a pure enzyme (Fig. 2). The sodium dodecyl sulfate gel of the DEAE-fraction displays three major and five minor bands. Only a single protein band remains after the 85" heat step, and this band migrates with a mobility slightly greater than a chymotrypsinogen marker.
The intensity of the corresponding band in the DEAE-fraction is consistent with a 7-fold purification by the 85" heat step. A single band in sodium dodecyl sulfate gels implies that the enzyme consists of subunits of identical molecular weight. The subunit molecular sorption of the enzyme to DEAE-Sephadex results in a 40 to Electrophoresis is performed as described under "Experimental Procedure." The samples are : A, carbonic anhydrase, 2 rg; B, brain DEAE-fraction, 4 pg; C, brain 85" supernatant fraction, 0.6 rg; D, same as C, but enzyme heated for an additional 10 min at 85"; E, chymotrypsinogen, 2 a. 50% loss in enzyme activity.
We have tested for the removal of essential factors by combining various fractions but have not been able to restore or to enhance enzyme activity.
The final step in the purification scheme depends on the high heat stability of enzyme first incubated in PRPP.
After heating DEAEfraction enzyme for 10 min at 85", 80% of enzyme activity remains and other proteins are denatured and precipitate.
The complete purification procedure can be conducted in 2 days and results in a 540-fold increase in enzyme specific activity.
The purified enzyme retains at least 60% of its activity after 4 months of storage in a liquid nitrogen freezer. Enzyme fractions through the DEAE step are not inactivated by repeated freezing and thawing.
We have purified hypoxanthine-guanine phosphoribosyltransferase from brain because the specific activity of brain extracts (17 units per g) is about 4 times that of liver (4.0 units per g) or cell extracts (4.2 units per g). An alternate procedure is described for the purification of enzyme from liver, and the purification procedure for enzyme from tissue culture cell extracts will be described in a subsequent paper. weight of hyposallthitle-guaIlille phosphoribosyltransferase determined by comparison with standards run on the same acrylamide slab gel is 25,000 (Fig. 3). Enzyme from liver or cells has the same subunit molecular weight.
The molecular weight of the native enzyme was determined by Sephades column chromatography and by acrylamide gel electrophoresis.
In gel filtration experiments with Sephades G-150 or G-100, hypoxanthine-guanine phosphoribosyltransferase activity elutes in a single peak. The native molecular weight of the enzyme determined by column chromatography on Sephadex G-100 (Fig. 4) is 78,000. An error of one fraction on either side of the assigned peak would give a molecular weight range of 73,000 to 83,000.
When run on non-denaturing acrylamide gels, the purified enzyme appears to consist of three or four bands (Fig. 5). Assay of gel slices indicates a broad peak of activity covering the area of these bands (Fig. 5). The slices \verc too wide to detect the possible existence of separate peaks of activity.
The native molecular weight was determined from the mobility of the enzyme in acrylamide gels of varying acrylamide concentration as described by Hedrick and Smith (19). The log of the mobility was plotted as a function of gel concentration for protein standards and for the enzyme bands on gels of 4.5, 6, 7.5, and 9y0 acrylamide.
The three enzyme bands which we could measure give a family of parallel lines (Fig. Sil) which implies that they have the same molecular weight but different net charges. The slopes of the plot of the log of the mobility versus gel concentrations are proportional to the molecular weight as shown in the standard curve in Fig. 6B. The enzyme native molecular weight determined from the standard curve is 85,000.
Electrofocusing of the brain DEAE-fraction gives three major peaks containing 22, 28, and 5074 of enzyme activity (Fig. 7) with isoelectric pH values of 6.24, 6.43, and 6.55, respectively. Experiments with liver DEAE-fraction also give three major peaks containing 8, 22, and 70cfi of enzyme activity with isoelectric pH values of 6.33, 6.49, and 6.70, respectively.
It is not clear whether the observed differences in isoelectric pH values and relative distribution between brain and liver enzymes rep- The void volume of the column is determined with blue dextran (+m), located by the absorbance of 6G0 nm. Hypoxanthine-guanine phosphoribosyltransferase (A---A) is located by the assay described under "Experimental Procedure," and bovine serum albumin, by the absorbance at 280 nm. Glucose-6-P dehydrogenase (n---n) and hexokinase (O--O) activities are located by assays (14, 15) measuring the increase in absorbance at 340 nm. The molecular weights of the standards are: bovine serum albumin, 68,000 (16) ; glucose-6-P dehydrogenase, 103,700 (17); yeast hexokinase, 48,000 at pH above neutral (18). resent real variation, or reflect differences in the methods of preparing the two enzymes.
In the presence of the substrate PRPP, hyposanthine-guanine phosphoribosyltransferase is extremely heat-stable. Incubating the enzyme with PRPP increases the activity 2.5.fold and protects the enzyme against heat inactivation, as shown in Table II. The protected enzyme retains 96 and 807; activity after 10 min at 65 and 85", respectively, while the unprotected enzyme retains only 66 and 19yc, activity at these two temperatures.
illeasurements of IMP formation as a function of time (not shown) indicate that the increased enzyme activity observed after incubation with PRPP is due to an increase in the rate of IMP synthesis.
Chinese hamster hypoxanthine-guanine phosphoribosyltransferase is active over a broad pH range, with optimum activity at pH 10 to 10.5. hs shown in Fig. 8, the activity increases about 2.5-fold as the pH changes from pH 5.6 to 7, is fairly constant from pH 7 to 8, and then increases 3.fold as the pH rises from 8 to 10.5. Beyond pH 10.5, the activity falls off sharply.
The effect of pH on enzyme activity may reflect the ionic state of the substrates.
PRPP has pK values of 5.9 and 6.7 (20), and hypoxanthine has a pK of 8.9. The location of enzyme activity, measured on an adjoining section of gel, is shown to the right. Standards (1 to 4 pg) and 2 pg of liver 85" supernatant are run on 4.5, 6, 7.5, and 9% acrylamide gels as described under "Experimental Procedure." The mobility of the bands is determined relative to the dye band. The slopes of the plots of log mobility versus gel concentration are plotted here as a function of molecular weight. The standards and their molecular weights are: carbonic anhydrase (29,000), ovalbumin (43,000), bovine serum albumin (SS,OOO), creatine kinase (81,000), and glucose-6-P dehydrogenase from Leuconostoc mesenteroides (103,700).
Hypoxanthine-guanine phosphoribosyltransferase displays Michaelis-Menten kinetics when one of the substrates is in limiting amounts and the other is in excess. Lineweaver-Burk plots of (velocity)+ versus (substrate)-' for PRPP, hypoxanthine and guanine are shown in Fig. 9: The velocities are expressed as V,,,,,/V so that the data for the enzyme from liver, brain, and cells can be plotted on the same scale. The enzyme from these three sources has identical K, values for a given substrate.
The apparent K, values for hypoxanthine and guanine Electrofocusing is performed as described under "Experimental Procedure" with brain DEAE-fraction. Enzyme activity (O-O) is measured as described under "Experimental Procedure. "   TABLE  II Heat inactivation of hypoxanthine-guanine phosphoribosyltransjerase A sample of liver DEAE-fraction containing 1.9 mg of protein per ml was incubated overnight at 0" with 1 mM PRPP.
The sample was then diluted 50-fold with enzyme buffer containing 1 mM PRPP. An untreated enzyme sample was diluted to the same concentration but without PRPP. Both samples were heated for 10 min at the temperatures indicated, cooled, centrifuged, and assayed for enzyme activity.   The activity of hypoxanthine-guaninc phosphoribosyltransferax is subject to inhibition by its product nucleot,ides 1RlP or GAII', but not by hhll', as shown in Table III. The inhibition by IlIP is neither strictly competitive with I'RPP, 1101' strictly noncompetitive.
At 1 mM IAII', the K, for PRPI' is increased to 28 ~.ll"r, and the V,,,,, is decreased to one-third of normal (data not shown).
No inhibit)ion is seen with ribosc-l-1' as high as 5 mM. The nucleoside inosine is slightly inhibitory (81 y0 of control activity at 5 mat inosine). soluble protein.
Based on a native molrrular wright of 80,000, each cell contains approximately 3 x IO" molecules of hypoxanthine-guanine phosphoribosyltransfcrasc.
Using this figure, one can calculate the potential contribution of the enzyme to the synthesis of purine nucleotidcs in cells. Under optimal conditions the enzyme in each cell could synthesize 0.2 pmolc of IMP per generation time of 12 hours. The average I>NA plus RNA content of an animal cell in culture is 40 pg (21), or about 0.06 pmolc of purine mononuclcotides.
Thus, a cell could synthesize by hyposant,hine and guanine salvage alone, about 3 times the rninirnum amount of purine nucleotides nceded to duplicate its DNA and RNA in one gcncration.
The capability of cultured Chinese hamster ~11s to supply their purine nucleotidc needs by salvage pathways can be investigated by blocking de novo purinc synthesis with the drug aminopterin. Under these conditions, cells show only a minor increase in generation time implying that hyposarlthirle-guarliiic phosphoribosyltransferasc must be operating at close to maximal efficiency. From these data me conclude that the concentration of PRYI' and hyposanthine in cells must approach or escced the K, values for the enzyme. Hyposanthinc is present in esccss (30 FM) in F12 tissue culture rnrdium.
JYe estimate that the intracellular concentration of PRI'P is 0.1 mM based on the rate of IMP formation by crude ccl1 estracts in the absence of added PRPI'.
Hyposanthinc-guanine phosphoribosyltransfcrasc from Chinese hamster brain, liver, and cultured cells appears identical in native molecular weight, subunit molecular weight, and kinetic properties.
13ased on the specific activities of tho crude extracts, the enzyme represents 0.2(yc of the soluble protein in brain, and O.O4ci, of the soluble protein in liver and in cells. An extract from 10' cultured cells contains approximately 1 mg of Table IV presents a comparison of hypor;anthirle-gua~line phosphoribosyltransfcrase isolated from Chinese hamsters, human red blood cells, rat brain, and brcwcrs' yeast. Although the specific activities of all of the enzymes arc approsimately the same, the K, values observed for PIW', hypoxanthinc, and guanine for the Chinese hamster enzyme arc all considerably lower than those reported for the enzyme from other sources. The most striking diffcrcnces occur in the molecular weight determinations. The human enzyme has a native molecular  (24). The composition of Chinese hamster hypoxanthine-guanine phosphoribosyltransferase appears to be quite different.
The native molecular weight of Chinese hamster hypoxanthineguanine phosphoribosyltransferase is 78,000 to 85,000. Sodium dodecyl sulfate gel electrophoresis gives a subunit molecular weight of 25,000. The simplest interpretation of these data is that Chinese hamster hypoxanthine-guanine phosphoribosyltransferase consists of three identical subunits of molecular weight 25,000 for a combined native molecular weight of 75,000. However, since the native molecular weight may be as high as 85,000, we cannot eliminate the possibility that there are additional subunits of combined molecular weight less than 10,000. Our current' techniques arc not sufficiently sensitive to detect a small amount of low molecular weight protein.
i2t least three apparent isozymes of hypoxanthine-guanine phosphoribosyltransferase are present in our preparations and can be distinguished either by polyacrylamide gel electrophoresis or by electrofocusing.
The irregular peak we observe on elution of the brain enzyme from DEAE-Scphadex may also indicate some separation of isozymes.
Similar isozymes have been described by Arnold and Kelley (24) for the human enzyme, although their isoelectric points are somewhat lower. All evidence indicates that there is only a single gene locus for hypoxanthine-guanine phosphoribosyltransferase (24). Therefore, the observed isozymes probably result from post-translational modifications.
The relative amounts of the three major isozymes distinguished by electrofocusing differ between brain and liver. These differences may indicate variation in the isozyme distribution in the two tissues, but could also result from differences in the preparation of the two enzymes.
The shape of the enzyme peak eluted from DEAE-Sephades varies among preparations, and the pooled fractions may not contain the same proportions of each isozyme.
Also, the liver enzyme was neither heated nor incubated with PRPP.
An interesting aspect of our results is the similarity in molecular weight and subunit composition of hypoxanthine-guanine phosphoribosyltransferase, which converts hypoxanthine to IMP in the presence of PRPP, and purine nucleoside phosphorylase, which converts hypoxanthine to inosine in the presence of ribose-1-P.
Purine nucleoside phosphorylase from calf spleen appears to be a trimer with a subunit molecular weight of 28,000 and a native molecular weight of 84,000 (27). The trimerit structure of human purine nucleoside phosphorylase is also suggested by the demonstration of three binding sites for hypoxanthine per enzyme molecular weight of 81,000 (28), and by in vitro hybridization of mouse and human enzymes (29). Davies and Dean (30) found three or four coincident peaks of hypoxanthine-guanine phosphoribosyltransferase and purine nucleoside phosphorylase activity on electrofocusing of human erythrocyte lysates, suggesting that the activity of the two enzymes may reside in the same protein.
We have examined the relationship between hypoxanthineguanine phosphoribosyltransferase and purine nucleoside phosphorylase from Chinese hamsters. Purified hypoxanthineguanine phosphoribosyltransferase fails to convert hypoxanthine to inosine in the presence of ribose-1-P.
Furthermore, the substrate and product of purine nucleoside phosphorylase, ribose-1-P and inosine, have a negligible effect on the activity of hypoxanthine-guanine phosphoribosyltransferase. Experiments in this laboratory indicate that Chinese hamster purine nucleoside phosphorylase is much more heat-labile than hypoxanthineguanine phosphoribosyltransferase.
Similar differences in heat stability have been reported for human purine nucleoside phosphorylase and hypoxanthine-guanine phosphoribosyltransferase (1). Yet, it is still possible that the two enzymes share common subunits.
We are presently investigating this possibility. We have characterized hypoxanthine-guanine phosphoribosyltransferase from Chinese hamster brain, liver, and tissue culture cells. The enzymes from all three sources appear identical in structural and catalytic properties.
Thus, we can be reasonably confident that any differences we observe in drug-resistant cells