[1] 10w Reconstitution of Diphtheria Toxin from Two Nontoxic Cross-Reacting Mutant Proteins

[1] 140w The isolation of a new type of mutant Corynephage /3, which carries a missense mutation in the structural gene for diphtheria toxin synthesis is described. The lysogenic C703197)tox-crm+ strain of Corynebacterium diphtheriae produces a nontoxic, extracellular protein of molecular weight 62,000. This protein is immunologically indistinguishable from toxin itself but inhibits the action of toxin on IfeLa cells, probably by competing for attachment sites on the cell membrane. In contrast to fragment A derived from diphtheria toxin, fragment A197 is unable to catalyze the inactivation of eucaryotic polypeptidyl-transfer RNA- transferase 11. When mixtures of the two nontoxic mutant proteins, enzymically active crm45 protein and enzymically inactive crm197 protein, are subjected to mild treatment with trypsin in the presence of a thiol and then allowed to reoxi- dize after dialysis to remove excess thiol, "diphtheria toxin" is reconstituted in high yield.

[1] 338w We have since isolated several other mutant strains of /3, each carrying an altered tox structural gene. Lysogenic C7 strains derived from certain of these mutant phages release cross-reacting proteins that are of reduced toxicity be- cause of alterations that affect enzyme activity in the NH2-terminal fragment A portion of the molecule. We now report on the properties of an extracellular protein released by C78 (/3197) tOx-Cr+ which, though completely nontoxic, ap- pears to be immunologically identical with intact toxin. Fragment A197 derived from crm197 has no enzyme activity, although the intact protein crm,97 can compete with diphtheria toxin for attachment sites on the HeLa cell membrane. We show that by suitable treatment of mixtures of crm45 (lacking the terminal sequence of fragment B, 17,000 daltons) and crm197 (with defective fragment A), a fully active toxic protein can be reconstituted in good yield. Phage mutants were -isolated by a modification of the method used to isolate 845 (5). Nitrosoguanidine was added 20 minutes after ultraviolet ir- radiation of cultures of C7,,(,); 3 to 3.5 hours later, the culture filtrates con- taining the surviving phage (0.6 to 1 percent of the normal burst) were plated on C7(-). Lysogenized resistant colonies were tested for toxinogeny by the rabbit intradermal test. Several liters of the strain C7.(/8197) tox-Crm+ isolated in this way were grown under optimal conditions for toxin production by C7,(,8). The crm197 protein was purified from the culture supernatant by am- monium sulfate precipitation, dialysis, and chromatography on diethylaminoethylcellulose. Some properties of purified crml97 are compared with those of purified crm45, toxin, and toxoid in Table 1. The chemical and physical properties of the crm197 protein are very similar to those of diphtheria toxin itself. The two pro- teins have the same molecular weight of 62,000, as determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Toxin and crmr97 are immunologically indistinguishable, giving curves that are superposable when tested by quantitative flocculation against a horse antitoxin. The specific toxicity of crm,97, however, is less than one-millionth that of diphtheria toxin.

[2] 79w As much as 0.5 mg (250 flocoulating doses, Lf) of purified crmI97 injected into guinea pigs failed to produce even a local reaction at the site of injection. Two rabbits were first injected with 0.2 mg of crmI97 in complete Freund adju- vant; 1 month later, a booster injection of 0.2 mg was given, and a week later 0.4 mg was given. (There were no toxic reactions.) One week afterward the antiserums gave positive ring tests with 100 Fig.

[3] 53w 1. Competition between crm1,n and toxin for HeLa cell binding sites. Washed 0 suspensions of growing HeLa cells were o \ / suspended in Eagle's medium containing Q x pt 2 percent fetal calf serum and then dis- tributed in 2-ml amounts in roller tubes C containing increasing amounts of purified $ 50

[4] 45w crnbn7. After 30 minutes rotation at 5 rev/ 050 min at 370C, diphtheria toxin (1 /Ag/ml; 0.5 00 j \Lf/ml) was added to all tubes, except for E£ certain control tubes. After incubation for .' \ 3.5 hours more, 100 IAl of r'C]leucine J --

[5] 147w (1 Ac/ml) was added to each tube. Cells o were harvested 3.5 hours later on Milli- Toxin (pg/ml) pore filters, washed first with Hanks salt Ratio (crm/toxin) as 100 percent. In the presence of 60 ,ug of crMn.7 alone, leucine incorporation was the same as in the control. The open circles show the effect of increasing ratios of crm,7 to toxin on ["C]leucine uptake plotted as percentage of the control uptake. The crosses represent a titration of toxin under the same experimental conditions. Incorporation of leucine in the presence of toxin (1 ,ug/ ml) alone was 23 percent of the control value (dashed line). The closed circle and triangle show leucine uptake in the presence of crm45 and purified toxoid in ratios to toxin of 40: 1 and 60: 1, respectively. tAbility to block re- versibly the inhibition by toxin of amino acid incorporation into HeLa cells.

[6] 40w tExpressed as percent of intradermal toxicity of purified toxin. § Expressed as percent of antibody precipitable from a horse antitoxin by toxin. purified toxin and contained 1 and 5 units of neutralizing antitoxin per milli- liter by rabbit skin test.

[7] 430w When purified crm197 is subjected to mild digestion with trypsin in the pres- ence of dithiothreitol (DTT), it breaks down, as does toxin, into two dissimnilar fragments of 24,000 and 38,000 dal- tons, respectively, which can be sep- arated by SDS-polyacrylamide gel electrophoresis. Unlike toxin, however, neither intact crm197 nor fragment A197 derived from it shows any enzymic activity. Under the usual test condi- tions (2), as little as 2 ng of activated diphtheria toxin catalyze the rapid breakdown of NAD and transfer of its ADP ribose group to eucaryotic trans- locase. Under identical conditions, 10,000 times as much crm197 failed to show detectable ADP-ribosylating ac- tivity, whether tested before or after treatment with varying amounts of tryp- A sin in the presence of DTT. The failure of crm197 to show toxicity is therefore due to a mutation causing loss of the enzymic activity associated with the fragment A portion of the toxin molecule. It seems likely that the alteration leading to loss of enzymic activity is located in the NAD-binding site. In the first place, crm197 protein has no effect on the enzymic activity of nicked toxin in vitro even when the ratio of crm197 to toxin exceeds 100: 1. This result shows that crm197 does not interfere with the formation of the ternary com- plex of fragment A, and NAD, and translocase (6). Finally, we have been unable to detect appreciable binding of [14CJNAD even by 1.5 percent solu- tions of purified crm,97, by the technic of equilibrium dialysis. Under the same conditions, binding of NAD by far low- er concentrations of toxin is easily mea- sured (7). Reconstitution of diphtheria toxin in mixtures of purified crm45 and crms97. A mixture was prepared in 0.05M tris at pH 8.0 containing crm4s (2.4 mg/ml) and crmino (1.6 mg/ml) equivalent to a molar ratio of 2: 1. A portion of the mixture was treated with crystalline trypsin (5 ug/ml) in the presence of 10 mM DTT. After 10 minutes at 370C, the reaction was stopped by addition of soy bean-trypsin inhibitor. The ADP- ribosylating activity of the mixture was increased about twofold by this treatment. The digestion mixture was then dialyzed in the cold against 0.01M phosphate, pH 7.2 to remove the excess DTT. Both mixtures, before (A) and after (B) treatment with trypsin, were analyzed in duplicate on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, with 10and 20-,ul samples, respectively. No thiol was added to the gels. One of each of the gels was stained (shown at top); the other was analyzed for ADP-ribosylation activity (2) after elution from 2-mm slices.

[8] 225w 90Q Table 1 also shows that the nontoxic crm45 protein, w,hich lacks the 17,000- dalton COOH-terminal amino acid se- quence of toxin, has very different prop- erties from those of crm197. Although fully active enzymically, crm45 is non- toxic because it cannot be taken up by sensitive cells. It was therefore neces- sary to find out whether the enzymically inactive crm197 still retained the capacity to interact with specific toxin- binding sites on the HeLa cell mem- brane. Figure 1 shows that this is indeed the case and that crm197 can success- fully compete with toxin for attachment to HeLa cells and thus block the inhibition by toxin of amino acid incor- poration into cell protein. The number of toxin-binding sites on the HeLa cell membrane is small (8). From the pre- liminary data, the affinity of crm197 for these sites would not appear to be very different from that of toxin itself. Nev- ertheless, even a large excess of purified crm45 or of purified diphtheria toxoid failed to interfere significantly with the binding of toxin by the cells. We con- clude from these experiments that the COOH-terminal, 17,000-dalton portion of fragment B is essential for attach- ment of toxin to the HeLa cell mem- brane and that in its absence enzymically active fragment A portion of the molecule cannot gain access to the cytoplasm.

[9] 28w Final and striking confirmation of the above conclusion has come from exper- iments in which toxin molecules were reassembled from fragments derived from the two nontoxic crm proteins.

[10] 155w When purified preparations of crm45 and crm197, containing only traces of nicked protein (1) were mixed and allowed to remain overnight in the cold, the mixtures became toxic, although the actual amount of toxicity developed was very low. However, when similar mix- tures were kept under the same condi- tions in the presence of 10mM DDT, appreciable toxicity always developed and was equivalent to about 2 to 5 percent of that to be expected if all of the crm,97 in the mixture had been con- verted to toxin. It 'is estimated that about 5 to 10 percent of the crrm197 in the purified preparation was present in the nicked form. These observations suggested that toxin was being recon- stituted from fragments of crm pro- teins. The yield of reconstituted toxin is greatly increased when the mixture of proteins is treated with trypsin in the presence of DDT (Fig. 2). Be- fore treatment, the nontoxic mixture

[11] 120w SCIENCE, VOL. 175 ._ 4-. gave two major bands on SDS-gel electrophoresis at positions corresponding to proteins of 62,000 daltons (crm197) and 45,000 daltons (crm45) (Fig. 2A). Almost all of the enzymic activity was associated with the crm45 peak, but it is evident that a small amount of free fragment A (24,000 daltons) was also present. After removal of DTT and reoxidation, a significant proportion of the activity had moved over into the 62,000-dalton peak (Fig. 2B). In fact, the shift of enzymic activity into this peak corresponds to nearly 25 percent of all the crm45 originally added and is equivalent to approximately half of the activity to be expected if all the crmL97 had been converted into active enzyme.

[1] 210w Diphtheria toxin is synthesized as a single polypeptide chain of about 62,000 daltons by Corynebacterium diphtheriae lysogenic for the temperature phage #3to'+. When purified toxin is treated with trypsin in the presence of a thiol, a single peptide bond is split (nicked), and the two disulfide bonds are reduced to yield two fragments, A (24,000 daltons) and B (38,000 daltons) (1); both fragments are required for toxicity (2). In the presence of nicotinamide adenine dinucleotide (NAD), fragment A cata- lyzes the adenosine diphosphate (ADP)ribosylation and inactivation of soluble eucaryotic polypeptidyl-transfer RNA- transferase II (translocase) (3). It has been suggested that the COOH-terminal fragment B portion of the toxin molecule is required for attachment to and penetration of the membrane of the sensitive cell (4,5). Uchida et al. (5) have reported the isolation of a tox- mutant phage, /345. After lysogenization of the sensitive C78(-)ttoxstrain with /345, the converted C7,(/345)t°x-crm+ strain produces a nontoxic, extracellular protein, crm45 (cross-reacting material), of molecular weight 45,000, that is enzymically active and cross-reacts with diphtheria antitoxin. After mild treat- ment of crm45 with trypsin in the presence of thiol, a 24,000-dalton frag- ment, indistinguishable from fragment A, is formed. The protein crm45 is nontoxic because its fragment A fails to reach the sensitive cell interior.

[2] 50w The trypsin-treated dialyzed mixture was injected, intraperitoneally, into guinea pigs (250 to 280 g) in amounts corresponding to 10, 5, and 2.5 ,ug of the crm197 originally present. The survival times were 10 to 12, 14 to 16, and 20 hours, respectively. From the dose-survival curve of Baseman et al.

[3] 49w (4), it may be calculated that the yield of reassembled toxin corresponds to between 40 and 80 percent of the crm197 originally added, a result in excellent agreement with the proportion of enzymic activity shifted into the 62,000-dalton peak (9). Antitoxin specifically neutralized the toxicity of the reconstituted mixture.

[4] 131w In conclusion, we have demonstrated that each of the two dissimilar fragments comprising the diphtheria toxin molecule has a separate and distinct function. Toxicity depends on the unique enzymic activity associated with fragment A. This activity is responsible for the arrest of polypeptide chain elongation in sensitive cells by specific ADP-ribosylation of translocase. But fragment A can only reach the sensitive cell cytoplasm when it is specifically associated with fragment B. The number of sites on HeLa cells available to react with toxin molecules is small (8). Since crmI97, but not crm45, suc- cessfully competes with toxin in HeLa cell cultures, it seems most probable that attachment and penetration of toxin are specific processes that depend on the composition and conformation of the 17,000-dalton COOH-terminal amino acid sequence of the molecule.

[5] 126w TsuyosHi UCHIDA A. M. PAPPENHEIMER, JR. ANNABEL AVERY HARPER Biological Laboratories, Harvard University, Cambridge, Massachusetts 02138 (despite the fact that crm, was in twofold excess), possibly because fragment A has a higher affinity for intact fragment B derived from crml,9 than for the corresponding 21,000- dalton fragment derived from crm1, or for another fragment A molecule to form a dimer. Because of the relative instability of fragment B, some degradation probably al- ways occurs, and the reconstituted mixture therefore contains an excess of fragment A, seen as a fast-moving component. 10. Aided by NIH grant 09006 and NSF grant GB18919. We thank Dr. R. Y. Gottshall of the Michigan Department of Health for con- centrated diphtheria toxin and Robin Greamy for technical assistance. 30 September 1971

[6] 69w A-bstract. Three isozymes of phenylalanine hydroxylase exist in adult rat liver. They are chromatographically unique. Partial chracterization suggests that they are similar in chemical properties and differ only in charge. Estimation of the Stokes radii indicates that the isozymes have similar molecular weights of about 200,000. Two isozymes exist in human fetal liver. Alterations of the relative amounts of these isozymes may control the phenotype of the disease phenylketonuria.

[7] 65w Phenylalanine hydroxylase (E.C. 1.14.3.1) catalyzes the irreversible hydroxylation of phenylalanine at the para position to yield tyrosine (1). The hydroxylating system consists of several protein components (2) and requires reduced nicotinamide adenine dinucleotide phosphate and pteridine as cofactors (3). Huzino and Bessman separated the hydroxylase component into two isozymes (4). Kaufman and Fisher also found two forms of the enzyme and further characterized them (5).

[8] 50w We have separated and character- ized three isozymes from adult rat liver and have distinguished between two isozymes from human fetal liver. The three forms isolated from rat liver, which we have called pi, kappa, and upsilon, appear to be three distinct isozymes of phenylalanine hydroxylase and not conformational isomers.

[9] 156w All procedures were carried out at 40C unless otherwise noted. Male rats (Sprague-Dawley strain), 150 to 250 g, were decapitated, and the livers were homogenized in two volumes of buffer that contained 0.025M K2HPO4, 0.1 5M KCl, and 5 mM dithiothreitol (D1i), pH 6.8. The homogenates were centrifuged at l00,OOOg for 30 minutes, and a volume of clear supernatant containing 30 mg of protein (estimated by the biuret method) was placed on a column of neutral calcium phosphate gel and wet cellulose powder (30:70, by volume). Column dimensions were 7 by 200 mm and bed vol- ume was 10 ml. The sample was eluted with a linear gradient of K2HPO4, 0.025 to 0.2M, pH 6.8. The gradient also contained 0.1 5M KCI and 5 mM DTT; total volume was 50 ml. Flow rate was approximately 0.25 ml/min, and all of the recoverable activity was eluted in 100 0.4-ml fractions. The yield was 90 to 100 percent.

[10] 24w The specific activity of the most active fraction of the kappa peak was about tenfold greater than the specific activity of the crude supematant.