Operon, , 2,56 with the C Gene the Operator Gene

VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 335

The authors are indebted to Mrs. J. H. Crookston for quantitation of the two red-cell popula- tions, to Dr. A. G. Steinberg for determining the Gm groups, and to Prof. C. Stern for critical review of the manuscript.

* Supported in part by grants from the National Science Foundation (G-14825) and the National Institutes of Health (2A-51901 and H-5780). t With the technical assistance of Barbara Burt. 1 Goldschmidt, R., and K. Katsuki, Biol. Zentralbl., 51, 58-74 (1931). 2 Whiting, P. W., and A. R. Whiting, Biol. Bull., 52, 89-117 (1927). 3 Hollander, W. F., J. Hered., 40, 271-277 (1949). 4Bierens de Haan, J. A., Roux' Arch. Entw. Mech., 37, 420-432 (1913). Mangold, O., and F. Seidel, Roux' Arch. Entw. Mech., 111, 593-665 (1927). 6 Nicholas, J. S., and B. V. Hall, J. Exptl. Zool., 90, 441-459 (1942). 7Owen, R. D., Science, 102, 400-401 (1945). 8 Dunsford, I., C. C. Bowley, A. M. Hutchison, J. S. Thompson, R. Sanger, and R. R. Race, Brit. Med. J., 2, 81 (1953). 9 Booth, P. B., G. Plaut, J. D. James, E. W. Ikin, P. Moores, R. Sanger, and R. R. Race, Brit. Med. J., 1, 1456-1458 (1957). 10 Nicholas, J. W., W. J. Jenkins, and W. L. Marsh, Brit. Med. J., 1, 1458-1460 (1957). "1 Moorhead, P. S., P. C. Nowell, W. J. Mellman, D. M. Battips, and D. A. Hungerford, Exptl. Cell Research, 20, 613-616 (1960). 2 Waxman, S. H., S. M. Gartler, and V. C. Kelley, J. Pediat. (in press). 13 Race, R. R., and R. Sanger, Blood Groups in Man, 3rd ed. (Oxford: Blackwell Scientific Publications, 1958).

DUAL EFFECTS OF STRUCTURAL GENES IN ESCHERICHIA COLI* BY NANCY LEEt AND ELLIS ENGLESBERG

DEPARTMENT OF BIOLOGICAL SCIENCES, UNIVERSITY OF PITTSBURGH Communicated by Milislav Demerec, January 3, 1962 L-arabinose isomerase, L-ribulokinase, and L-ribulose 5-phosphate 4-epimerase are the first three enzymes involved in the sequential breakdown of L-arabinose by Escherichia coli, strain B/r.'. 2 Twenty-two L-arabinose negative mutants have been isolated and shown to be closely linked in a linear order between the markers threonine (thr) and leucine (leu) and divisible into four functional and genetically discrete groups or genes (A, B, C, D) .1-4 Gene A mutants are deficient in the enzyme L-arabinose isomerase and have higher L-ribulokinase activity than the wild type. Gene B mutants are deficient in L-ribulokinase activity and have either increased or decreased inducible levels of both L-arabinose isomerase and L-ribulose 5-phosphate 4-epimerase activities. Gene C mutants are deficient in all three enzymes. Gene D mutants are deficient in L-ribulose 5-phosphate 4-epimerase" 2 (see Fig. 1). Since L-arabinose is apparently the inducer of all three enzymes, it has been proposed that this entire arabinose region probably represents an operon, , 2,56 with the C gene the operator gene,' and genes A, B, and D the structural genes for L-arabinose isomerase,I L-ribulokinase,' and L-ribulose 5- phosphate 4-epimerase,2 respectively. This paper will present evidence that the A and B genes are indeed the structural Downloaded by guest on September 26, 2021 336 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

genes for L-arabinose isomerase and L-ribulokinase respectively, that increased amounts of wild-type L-ribulokinase are produced by mutation within the gene A, and that increased or decreased amounts of wild-type L-arabinose isomerase are produced by mutation within the gene B.

B~~~~~~ inr139 53 2 13 i3a 7 4 6 6 23 all 15 i27 24 8 2 5 3 ~~~~~~~~~~~2119 128 leu

CH -

-12H 1 HOO10H ___ Grabinose______HO H L r.2 i0K rPu'Er-10 r buiose_5-prloSphote H orerosrHaO -emrse H--O H0r-1 O~-101PO-K ep eras H 2~~~~~~~~~~~

FIG. 1 E coli L-arabinose gene-enzyme complex or operon.

Materials and Methods.-E. coli, strain B/r wild type, and mutant strains, the method of preparation of cell-free extracts, and enzymatic assays for L-arabinose isomerase and L-ribulokinase have been described.', Besides the manometric method of assaying the enzyme L-ribulokinase,l a spectrophotometric method was also employed. The reaction mixture contained: DPNH (Calbiochem), 0.3 mg; lactic dehydrogenase (Worthington rabbit muscle, containing pyruvate kinase), 0.25 mg; reduced glutathione (Schwarz), 2 MM; EDTA, 1 MM; MgCl2, 10 JAM; phosphoenolpyruvate (Cal- biochem trisodium salt dihydrate), 5 MM; ATP (Schwarz, ATN.), 2 JAM; L-ribulose (prepared by the method of Englesberg'), 5 MM; Tris, pH 7.6, 90 MM; and cell extract, in a total volume of 3.0 ml. This mixture was placed in a quartz cuvette of 1 cm light path, and L-ribulokinase activity was determined by measuring the disappearance of DPNH at 340 mj~at 30-sec intervals for 3 min at 370 and subtracting DPNH oxidation determined in the absence of substrate. L-ribulokinase activity was rate-limiting, up to 10 units of enzyme per reaction mixture. A unit of L-ribulokinase is the amount of enzyme which will phosphorylate 1 MM of L-ribulose per hr. Kinetics of the differential synthesis of L-arabinose isomerase and L-ribulokinase were determined as follows: 1,400 ml of casein hydrolyzate medium' was inoculated with 28 ml of an over- night nutrient broth culture, distributed into seven one-liter flasks, and incubated at 370 with shaking. When a cell density of about 150 Klett Summerson units was reached, L-arabinose was added to the cultures to yield a 0.4% solution. At suitable time intervals, flasks were removed, chloromycetin succinate added to a final concentration of 20 /. grams per ml, and the sample immediately chilled in ice. L-arabinose isomerase and L-ribulokinase activities were determined using cell-free extracts as previously described.'1 Dry weight of the cells per ml of culture was determined on the basis that 1 Klett unit = 1.38 X 10 6 gm dry weight bacteria per ml. Antisera were prepared from rabbits using enzyme preparations of wild-type E. coli, strain B/r. One group of rabbits was immunized with purified wild-type L-arabinose isomerase prepared by the DEAE cellulose fractionation method. This antigen preparation was free of L- ribulokinase activity (less than 0.02 unit per mg protein). When varying dilutions of this antigen were titrated by the Ouchterlony method' against an antiserum prepared against induced wild- type cell extracts, and containing antibodies against at least three enzymes of the L-arabinose pathway, only a single precipitin band was formed, whereas three distinct bands were formed with the crude extract. Each rabbit in this group received 1,065 units of L-arabinose isomerase in 8.25mg of protein. Another group of rabbits was immunized with wild-type partially purified extract (Table 1, step 3), which contained both L-arabinose isomerase and L-ribulokinase. Each rabbit in this group received 2,760 units of L-arabinose isomerase and 846 units of Lribulokinase in 52.9mg of protein.'3 Downloaded by guest on September 26, 2021 VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 337

TABLE 1 SEPARATION OF WILD-TYPE L-ARABINOSE ISOMERASE FREE OF L-RIBULOKINASE ,L-Arabinose Isomerase , -L L-Ribulokinase----- Total Over-all Over-all volume Total Specific % Total Specific % Steps (ml) units* activityt recovery units* activityt recovery Preliminary Purification 1. Crude extract 95.0 47,600** 7.16** 20,300** 3.07** 2. Dialyzed Mn supernatant 84.6 69,500 32.2 100 30,100 14.0 100 3. Heated dialyzed Mn supernatant 80.5 67,200 50.6 97 23,200 17.4 71 DEAE Cellulose Separation of L-Arabinose Isomerase and L-Ribulokinase, pH = 6.5 Heated Dialyzed Mn Supernatant placed on column 5.0 4,110 50.6 1,440 17.4 4. Total enzyme eluded from column 2,800 65 1,500 78 Pooled Enzyme Fractions: 5a. L-arabinose isomerase fraction 52 2,212 51 0 (#41-50) 5b. L-ribulokinase fraction 51 205 1,061. 55 (#55-62) Pooled Fractions, Concentrated by (NH4)2SO4 Precipitation 6a. L-arabinose isomerase fraction 5.0 1,300 130 30 0 6b. L-ribulokinase fraction 5.0 35 689 68.9 37 * Imoles of Lribulose produced per hr.' t units/mg protein.' + smoles of L-ribulose 5-phosphate produced per hr.' ** Estimated as a minimal value.

Before injections, all rabbits were bled for base-line sera. The antigens were prepared in Freund's incomplete adjuvant, 1 part of antigen to 1 part of adjuvant, the latter prepared by mixing 15% Arlacel A with 85% Bayol F. The first injection consisted of 2.0 ml, given in the hind food pads and between the scapula. A booster injection of 2.5 ml was given after 3 weeks, and one of 3 ml after 5 weeks. After each injection, the rabbits were bled for titer, and all animals were ex- sanguinated after 8 weeks. Agar gel diffusion plates (Ouchterlony) were prepared as follows: 0.6% Oxoid Ionagar #2 (Oxo Limited, London) and 0.8% NaCl were sterilized by autoclaving, aqueous merthiolate was added to a final concentration of 1: 10,000, and the mixture brought to pH 7 with HCl. The plates were poured and cut with a Feinberg #1801 cutter (Shandon Scientific Co., London). DEAE cellulose (Selectacel, Reagent Grade), before being used, was washed 10 times with an equal volume of 1 N NaOH, neutralized with phosphoric acid, washed again with 0.1 M phosphate buffer, and finally equilibrated on the column at 40 with 0.01 M phosphate buffer containing 10-3 M EDTA and 10-3 M reduced glutathione. The latter buffer will hence be referred to as the column buffer. Results.-Part 1. Demonstration of A and B loci as the structural genes for L-arabinose isomerase and L-ribulokinase respectively: Mutants in the A locus (ara-2, ara4, ara-7, and ara-13) are deficient in L-arabinose isomerase activity, and all mutants in the B locus (ara-1, ara-6, ara-8, ara-14, ara-15, ara-16, ara-23, and ara-24) are deficient in L-ribulokinase activity.' If the A and B loci represent structural genes, one might expect some A mutants to produce inactive proteins antigenically related to L-arabinose isomerase and some B mutants to produce inactive proteins antigenically related to L-ribulokinase. Furthermore, the production of these antigenically related proteins, or CRM's, should be under the same regulatory control mechanisms as are the corresponding wild-type enzymes. Downloaded by guest on September 26, 2021 338 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

In order to test for the presence of CRM's in mutant extracts, specific antisera were needed. For this purpose, an attempt was made to isolate L-arabinose isomerase and L-ribulokinase from wild-type extracts for use as antigens in the immunization of rabbits. Separation of wild-type L-arabinose isomerase free of L-ribulokinase: (1) Prepara- tion of crude extract: Wild-type E. coli B/r crude extract was prepared as follows. Seven liters of casein hydrolyzate medium was inoculated with 140 ml of an over- night nutrient-broth culturf and distributed in liter Erlenmeyer flasks in 500 ml quantities. The flasks were placed on rotary shakers at 370 until a Klett Summer- son reading of 250 (blue filter against culture medium blank) was reached. L-arabinose was added to a final concentration of 0.4%o and incubation continued for an additional 3 hr. The cells were washed and harvested as described.' The pellets were resuspended in 80 ml of EDTA 10-3 II, reduced glutathione 10-3 M solution, pH 7.4 (EDTA-glutathione solution), and treated in a Raytheon 10 kc sonic os- cillator in 10 ml quantities for 6 min. The combined material was then centrifuged at 10,000 rpm for 30 min and the supernatant collected. All high-speed centrifu- gations were performed using the 40 head of the Spinco refrigerated centrifuge. (2) Preliminary purification procedure: (See Table 1). To the above crude extract was added 5 ml (1/20th volume) of 1.0 M MnCl2, and the precipitate formed in 30 min at 00 was removed by centrifugation at 20,000 rpm for 45 min. The supernatant solution was dialyzed at 00 against three changes of 50 volumes of EDTA-glutathione solution for a total of 12 hr, and further clarified by centrifugation at 20,000 rpm for 45 min. Ten ml quantities of the Mn treated and dialyzed supernatant were heated to 530 in a water bath for 16 min, brought to 00, and centrifuged at 20,000 rpm for 45 min. (3) DEAE cellulose chromatography: The separation of L-arabinose isomerase from L-ribulokinase is accomplished by DEAE cellulose chromatography using a IVCl gradient and is pH-dependent (Fig. 2). Five ml of the heated material was dialyzed against 0.01 M phosphate buffer pH 6.5 in 10-3 M EDTA, 10-3 M reduced glutathione (column buffer) for 2 hr, and then placed on a DEAE cellulose column, 25 mm in diameter, containing 1 ml of gravity-packed DEAE cellulose per mg of protein chromatographed. The column was washed with '/2 column volume of buffer and eluted using a NaCl gradient from 0-0.22 111 at the rate of 1 ml per min. Five to seven ml fractions were collected and assayed for protein, L-arabinose isomerase, and L-ribulokinase. The eluates which contained L-arabinose isomerase but no L-ribulokinase activity were pooled, and solid ammonium sulfate was added at 00 to 95 per cent saturation. After 30 min of stirring, the precipitate was collected by centrifugation, resuspended in EDTA-glutathione solution, neutralized with KOH, and dialyzed against EDTA-glutathione solution at 00 until a negative test for SO4-- was obtained. This material will be referred to as the wild-type DEAE L-arabinose isomerase fraction (Table 1, step 6a). Lyophilization followed by dialysis can replace the ammonium sulfate concentration, with approximately the same recovery. For the immunization of rabbits, L-arabinose isomerase was prepared by the procedure above except that larger columns, 45 mm in diameter, were used to chromatograph 40 ml of extract each time. The recovery of enzyme from these large columns was lower.'4 For characteristics of the L-arabinose isomerase antigen Downloaded by guest on September 26, 2021 VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 339

pH of Column Buffer = 7.4

20- - 0.2 L-Ribulokinose "A L-arobinose Isomerose 10- Protein.

pH of Column Buffer =62 L-orobinose Isomerose > 20- L-Ribulkinme 0.20 FIG. 2.-Separation of wild-type Larabinose / Y isomerase and Lribulokinase on a DEAE cellu- X lose column as a function of pH. Protein was f l \ \ 0.1 determined as O.D.,80. Protein

pH of Column Buffer-6.5

40- 0.4 \ ,¶ L-Ribulokinose Protein 20- 0.2

LLorobinose u.. Isomerose A x_,_K 200 250 300 350 400 ml Volume Eluate

preparation, see Materials and Methods. Because of the trailing of L-arabinose isomerase into the L-ribulokinase peak in DEAE cellulose fractionation, it was not possible to prepare an L-ribulokinase free of L-arabinose isomerase by this procedure. Preparation of specific L-arabinose isomerase antiserum: L-arabinose isomerase purified by DEAE cellulose fractionation was used for immunization. The antiserum produced was absorbed twice for 30 min at 37° with an extract of E. coli B/r prepared from cells grown in the absence of L-arabinose. Such extracts are deficient in enzymes involved in L-arabinose metabolism. The amount of extract used each time was determined by pilot titrations to assure maximum precipitation without entering the zone of soluble complex formation. The absorbed serum was then heated to 560 for 30 min and clarified by centrifugation. Completeness of absorption was determined by the agar gel technique (Ouchterlony method) using various dilutions of uninduced cell extract.- All sera showed no band formation at the end of two weeks at 40 when tested against five twofold dilutions of uninduced cell extract. This serum did show a single precipitin band in two days at 40 when tested by the Ouchterlony method against crude extracts of L-arabinose induced wild-type cells. Furthermore, when tested in a similar manner with extracts of L-arabinose induced ara-2, which is deficient in L-arabinose isomerase but contains L-ribulo- kinasel and L-ribulose 5-phosphate 4-epimerase2 as well as other soluble antigens, no precipitin band was formed in two weeks. (This indicates that ara-2 does not Downloaded by guest on September 26, 2021 340 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

produce any L-arabinose isomerase CRM.) Thus, this antiserum is not able to precipitate proteins other than L-arabinose isomerase. (Although this antiserum precipitates L-arabinose isomerase, it does not destroy its activity.) The titer of the antiserum was determined by reacting a constant quantity of antiserum with varying amounts of wild-type L-arabinose isomerase. After 3 hr incubation at 00 the precipitates were removed by centrifugation at 12,000 rpm for 20 min, and the supernatants were assayed for remaining L-arabinose isomerase activities. A unit of anti-L-arabinose-isomerase is the amount of antibody which precipitates one unit of L-arabino-e isomerase at the equivalence zone. Anti-L-arabinose isomerase serum was standardized in batches and was apparently stable when frozen for periods up to a year. The particular batch of serum used in this study had a titer of 420 units/ml. Preparation and characterization of anti-L-ribulokinase serum: Antisera from rabbits immunized with a partially purified wild-type induced cell extract (Table 1, step 3) were absorbed with E. coli B/r extract prepared from uninduced cells as previously described for the preparation of anti-L-arabinose isomerase. This absorbed serum contained ATPase and DPNH-oxidase which interfered with spectrophotometric L-rubulokinase assays. To remove these impurities, the serum was first heated to 56 for 30 min and clarified by centrifugation. The supernatant was brought to 50 per cent ammonium sulfate saturation at room temperature, centrifuged, and the precipitate was resuspended in distilled water and dialyzed against 0.8% NaCl at 4°. This serum, completely absorbed with uninduced wild-type extract, formed three precipitin bands in the agar diffusion test with extracts of L-arabinose induced wild type, indicating the presence of three distinct antibodies to the inducible proteins involved in L-arabinose metabolism. The nature of the three bands was determined in the following manner. When this antiserum was further absorbed with purified wild-type L-arabinose isomerase, one of the three bands was no longer formed. Upon further absorption with a purified L-ribulokinase preparation8 prepared from ara-2, another band, apparently due to L-ribulokinase, failed to form. The remaining band was formed only by L-arabinose induced extracts of wild type, all group A and B mutants, and one D mutant tested, but was conspicuously absent in all C group mutants, indicating its association with the L-arabinose operon. In spite of the multiple specificity of this antiserum, it was possible to use this serum for the detection of L-ribulokinase CRM's. Using the agar gel diffusion technique with purified L-arabinose isomerase (wild-type DEAE fractionated) and purified L-ribulokinase (ara-2 DEAE fractionated), it was very simple to identify any un- known serologically active material by placing it on the same agar plate with the known enzymes. As would be expected from the agar diffusion test, the presence of the two other antibodies in this serum does not interfere with the L-ribulokinase-antikinase competitive precipitin reaction for the detection of L-ribulokinase CRM in crude extracts. The precipitation of L-ribulokinase was independent of concomitant precipitation of L-arabinose isomerase and other proteins (Fig. 3), since the number of L-ribulokinase units precipitated by a given quantity of antiserum at the equivalence zone was the same, whether other precipitating systems were present or not. The antiserum was standardized with respect to L-ribulokinase precipitating Downloaded by guest on September 26, 2021 VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 341

15 FIG. 3.-Independence of Lribulokinase c enzyme-antienzyme precipitin system from a Purified others in L-arabinose metabolism. A 1, 10- L-ribub1nase / constant quantity of polyvalent antiserum , against enzymes in L-arabinose metabolism X . is titrated with different amounts of purified L-ribulokinase from ara-2 (dotted line) and a '5- / whole extract (solid line) prepared from c E EX*OCt wild-type Larabinose induced cells. L- ' ribulokinase activity was determined spec- 04 trophotometrically. 0 6l 9 2 5 2l 2 o 3 6 91 15 I's1 21 24 27 30 Units L-Ribukinase Added

antibody by adding, to a constant quantity of antiserum, varying dilutions of a partially purified extract of induced wild-type cells (dialyzed Mn supernatant), allowing precipitation to occur at 00 for 3 hr, and assaying the amount of L-ribulokinase remaining in the supernatant. The titer of anti-L-ribulokinase antibody is the number of units of L-ribulokinase precipitated by 1 ml of serum at the equivalence zone. Unlike L-arabinose isomerase, L-ribulokinase was partially neutralized by the antibody. L-arabinose isomerase CRM and L-ribulokinase CRM-Inducibility of CRM's: A competitive precipitin test and the agar gel diffusion method of Ouchterlony were both used for the detection of CRM's. In the former procedure, the material to be tested is incubated with a known quantity of active enzyme and antienzyme. If CRM is present, more active enzyme will remain in the supernatant, since the competition between enzyme and CRM for antibody sites results in a protection of the enzyme from precipitation. Table 2 shows that, of the four A mutants tested with specific anti-L-arabinose isomerase serum (ara-2, ara-4, ara-7, and ara-13), the first three produced no detectable L-arabinose isomerase CRM, whereas ara-13 produced sufficient CRM to protect all the indicator enzyme present, although ara-13 itself contained no detectable amount of L-arabinose isomerase under the conditions of this experiment (Table 2, experiment #2, tube #13). Table 2 also shows that this CRM was present only in an induced extract of ara-13. The presence of L-arabinose isomerase CRM in ara-13 induced cell extracts was confirmed by Ouchterlony tests. Specific anti-L-arabinose isomerase serum reacted with ara-13 induced cell extract to form a precipitin band which merged with the band formed by the wild-type DEAE L-arabinose isomerase fraction. No precipitin band was formed by the uninduced extract of ara-13, nor was any precipitin band formed by ara-2, ara-4, and ara-7, induced or uninduced. Table 3 shows that all three group B mutants tested produced L-ribulokinase CRM. Control tubes showed that these B mutant extracts contained small amounts of L-ribulokinase in quantities insufficient to account for the degree of protection. The results were confirmed by Ouchterlony tests. The presence of L-ribulokinase CRM was found also in five other B mutants (ara-1, ara-6, ara-8, ara-15, and ara-23) by this method. No precipitin bands were formed when uninduced cell extracts were tested. The presence of L-arabinose induced, antigenically cross-reacting material (CRM) to L-arabinose isomerase in ara-13, a group A mutant, and to L-ribulokinase in Downloaded by guest on September 26, 2021 342 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

TABLE 2 THE DETECTION OF L-ARABINOSE ISOMERASE CRM IN GROUP A MUTANTS BY THE COMPETITIVE PRECIPITIN TEST Anti-isomerase Indicator CRM test Isomerase remaining Experiment Tube (units) isomerase (units) material (ml) in supernatant (units) 1 1 20 40 0 20.8 2 " 35 0 17.6 3 " 30 0 7.1 4 " 25 0 5.8 5 20 0 0 6 15 0 0 7 20 40 0.05 ara-4 induced 28.5 8 35 21.6 9 30 10.8 10 25 2.4 11 20 0 12 15 0 13 0 0.07 14 20 40 0.05 ara-7 induced 25.9 15 " 35 21.9 16 " 30 12.0 17 " 25 3.0 18 20 0 19 15 0 20 0 0.14 2 1 45 45.0 0 0 2 " 37.5 0 0 3 " 30.0 0 0 4 22.5 0 0 5 15.0 0 0 6 " 7.5 0 0 7 45 45.0 0.1 ara-13 induced 39.1 8 " 37.5 34.1 9 it 30.0 29.8 10 " 22.5 17.5 11 " 15.0 " 15.1 12 " 7.5 " 7.7 13 0 " 0 14 45 45.0 0.1 ara-13 uninduced 1.3 15 " 37.5 " 0 16 " 30.0 " 0 17 " 22.5 " 0 18 " 15.0 " 0 19 " 7.5 " 0 20 - 0 " 0 21 45 45.0 0.1 ara-2 induced 1.9 22 37.5 " 0 23 30.0 " 0 24 " 22.5 " 0 25 " 15.0 " 0 26 " 7.5 " 0 27 - 0 " 1.0 Specific anti-L-arabinose isomerase serum was prepared and processed as described in text. Diluted antiserum, CRM test material, and finally indicator L-arabinose isomerase were placed in small plastic centrifuge tubes. When necessary, sera were diluted with 0.8% NaCl, and enzymes were diluted with EDTA-glutathione solution. Total volume: 0.6 ml. In control tubes where no CRM test material was added, the volume was made up with bovine plasma albumin (25 mg/ml). The tubes were incubated at 00 for 3 hr and centrifuged at 12,000 rpm for 20 min, and the supernatants were assayed for remaining L-arabinose isomerase activity. Indicator enzymes were all freshly prepared and assayed. All CRM test materials were crude extracts. Units of enzyme and antienzyme were defined in text. "Induced" refers to extracts prepared from cells grown in a casein hydrolyzate L-arabinose medium; "uninduced" refers to extracts prepared from cells grown in casein hydrolyzate. group B mutants, indicates that the A and B loci are the structural genes for L-arabinose isomerase and L-ribulokinase respectively. Part 2. Studies on the dual effects in A and B loci. The L-arabinose isomerase of group B mutants: The differential rate of L-arabinose isomerase synthesis by the wild type was compared to that of a mutant which has the highest (ara-23) and one Downloaded by guest on September 26, 2021 VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 343

TABLE 3 THE DETECTION OF L-RIBULOKINASE CRM IN GROUP B MUTANTS BY THE COMPETITIVE PRECIPITIN TEST Tube Anti-ribulo- Indicator CRM test Ribulokinase remaining i kinase (units) ribulokinase (units) material (ml) in supernatant (units) 1 12 25 0 18.1 2 " 20 0 9.8 3 " 15 0 2.8 4 12 25 0.1 ara-14induced 30.5 5 " 20 " 23.6 6 " 15 " 11.8 7 12 25 0.1 ara-16 induced 31.3 8 " 20 " 25.0 9 " 15 " 15.3 10 12 25 0.1 ara-24 induced 27.8 11 " 20 " 29.2 12 " 15 " 15.3 13 0 0 0.1 ara-14 induced 1.76 14 0 0 0.1 ara-16 induced 1.83 15 0 0 0.1 ara-24 induced 3.96 See legend under Table 2. Conditions of test same as that for L-arabinose isomerase CRM, except that all CRM test materials were dialyzed Mn supernatants. Kinase assays were performed using the spectrophotometric method described in Materials and Methods. which has the lowest (ara-14) L-arabinose isomerase activity among mutants in the B gene (Fig. 4). With all three strains, the maximum differential rate of synthesis starts within 5 min after the addition of L-arabinose, and remains constant thereafter. The differential rate of L-arabinose isomerase synthesis by mutant ara-23 is 4.8 times that of the wild type; while with mutant ara-14, it is approximately 4.5 times less than the wild type. These values are in general agreement with the differences previously noted among these strains.1

I aro-23 I

e FIG. 4.-Differential enzyme synthesis of L-ara- 7 binose isomerase by wild-type and two B gene mutants, ara-23 and ara-14. L-arabinose was added to the wild-type culture at 0.22 mg dry wt bacteria/ml ; I WT to ara-23 at 0.21 mg dry wt bacteria/ml; and to 1 / ara-14 at 0.25 mg dry wt bacteria/ml.

Q2 02503 mg dry wt bacteria/ml culture

The L-arabinose isomerase assay reaction mixture was buffered at various pH's with 0.3 M Tris maleate from pH 6.5 to 7.9. The rates of reaction were determined with crude extracts of wild type, ara-23, and ara-14. The optimum pH was found to be 7.5 for all three. The experiment was repeated with DEAE cellulose fractionated Larabinose isomerase from wild type and from ara-23, the enzyme level Downloaded by guest on September 26, 2021 344 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

of ara-14 being too low to undergo this purification. Enzyme activity was absent at pH 6.0, rising to a maximum plateau at pH 7.8 to 8.1. Identical curves were obtained with wild-type and ara-23 enzymes. A variety of sugars were tested as substrates for the enzymes of wild type, ara-23, and ara-14, including L-arabinose, D-lyxose, D-ribose, D-mannose, D-glu- cose, D-galactose, and L-rhamnose. Of these sugars, there was detectable activity with only L-arabinose. K8 values for L-arabinose for the wild type (2.0 X 10-1 M), ara-23 (1.8 X 10-1 M), and ara-14 (1.9 X 10-1 M) are not significantly different (Fig. 5). The determination was again made with purified L-arabinose isomerase (DEAE L-arabinose isomerase fractions) from wild type and ara-23, with essentially the same results.

08-

0.7- ,

0.6- 0° > / . FIG. 5.-The effect of L-arabinose con- >0.5- ,, ,.,centration on the reaction velocity of L-ara- . / binose isomerase of the wild-type and two B gene mutants, ara-14 and ara-23. Reac- .020 tion mixture as previously described.' 04- Reaction velocity was measured as uM ribulose formed per 4 min for wild-type and ara-23 determinations, and per 30 min for 0.3 7 ara-14.

0.2 O ,) O/, 2-~~~~ 2 4 6 8 0o L-orobinose, M x 10 (S) The ratio of L-arabinose isomerase activity units to antigen units was approximately 1 for all B mutants tested, indicating that there is a corresponding increase or decrease in the amount of antigen over the wide range of enzyme activity ex- hibited by these strains (Table 4). It thus appears that the L-arabinose isomerase of B mutants has the same properties as the wild-type enzyme, and the difference in the level of enzyme activity is associated with increased or decreased amounts of enzyme protein. The L-ribulokinase of group A mutants: Kinetics of differential enzyme synthesis was determined in the same manner as described for L-arabinose isomerase. Figure 6 shows that the differential rate of L-ribulokinase synthesis in ara-2 is four times that of the wild type, which agrees with the differences in the specific activities of their extracts.' Downloaded by guest on September 26, 2021 VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 345

TABLE 4 COMPARISON OF ENZYME AcTIvITr UNITS To ANTIGEN UNITS OF WILD-TYPE AND A AND B MUTANT EXTRACTS Enzyme Activity Antigenicity L-arabinose isomerase Units of anti-L-arabinose Extract Preparation units per ml isomerase ppt'd per ml Wild-type heated dialyzed Mn treated extract 420 420* ara-14 crude extract 8.09 7.43 ara-1 crude extract 56.3 54.1 ara-23 DEAE L-arabinose isomerase fraction 971 921 L-ribulokinase Units of anti-L-ribulokinase units per ml ppt'd per ml Wild-type heated dialyzed Mn treated extract 224 224* ara-2 similarly treated 1,100 880 ara-4 similarly treated 359 383 ara-7 similarly treated 300 240 * By definition. Antisera were standardized as described in the text. To a known quantity of antienzyme was added the mutant extract containing a slight excess of enzyme, and after 3 hr incubation at 00 the supernatants were assayed for remaining activity.

There was no difference in the substrate specificities of wild-type DEAE fractionated L-ribulokinase (Table 1, step 6b) and a similar preparation of ara-2.8 Employing the spectrophotometric assay method, using 5 IAM of substrate, we found both enzymes were active on I-ribulose only and were inactive with respect to L-xylulose, D-xylulose, D-ribose, D-fructose, and L-arabitol. Both enzymes show similar K, values for L-ribulose, 1.7 X 10-4 M (Fig. 7), and for ATP, 0.87 X 10-4 M (wild type), 0.95 X 10-4 M (ara-2) (Fig. 8). The optimum pH of the wild-type and ara-2 L-ribulokinase was determined using the manometric technique.1 The reaction mixtures were buffered between pH 6.5 and 8.0 with NaHCO3 + C02. Both enzymes show identical pH curves with maximal activity at pH 7.7; there was no activity at pH below 7.

20-

FIG. 6.-D~iferetial synthesi of L-rbulokinseby wild- e07 type and an A gene mutant, ara-2. L-arabinose was added °/, /~~~~~~ to the wild-ty'pe culture at 0.44 mg dry wt bacteria/ml; S and to ara-2 at 0.36 mg dry wt bacteria/ml. > / ! /~~~~~

5- -

i0/

0.3 04; 0.5 0.6 0. mag dry wt. bocterio/m cu;ture Downloaded by guest on September 26, 2021 346 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

0.04 _ra-2

0.03- ,_, = 6 FIG. 7.-The effect of L-ribulose concentration on >,/ the reaction velocity of L-ribulokinase of the wild-type WT. Iand02-' of an A gene mutant, 0,02 ara-2. Reaction mixture as . 6/,Sdescribed/ in Materials and o /100--2 Methods. Approximately 2 _r,/,/l units/o' of enzyme were added. V Reaction was meas- 0.01- velocity | ;__2uredW oasI/Sthe2decrease=¢ in O.D.340 50- per 30 sec.

L-ribulose, M x le (S) Antigenic studies indicate that the increased levels of L-ribulokinase among the A mutants are accompanied by corresponding increases in the amount of L-ribulokinase protein (Table 4). It thus appears that the different A mutants produce different amounts of wild-type L-ribolukinase. Discussion.-Mutation in the A gene resulting in deficiency in L-arabinose isomerase causes the production (in one mutant out of four tested) of a protein (CRM) which still reacts with specific anti-isomerase serum. Mutation in the B gene resulting in deficiency in L-ribulokinase causes the production (in each of the mutants in the B gene) of proteins (CRM's), which still react with anti-ribulokinase. The CRM's, all of which have little or no enzyme activity, are produced by the A and B mutants only when they are grown in the presence of L-arabinose. This evidence indicates that genes A and B are the structural genes for L-arabinose isomerase and L-ribulokinase respectively. Mutations in these structural genes have a dual effect. Mutation in the L-arabinose isomerase structural gene, besides causing a deficiency in L-arabinose isomerase,

0.06

0.05-

-Ioro-2 FIG. 8.-The effect of ATP 5 ,>/concentration/ I on the reaction CQ03 C velocity of L-ribulokinase of the wild-type and of an A 0.02//XX50 WT gene mutant, ara-2. Condi- 00 l/ | v [ by/ |tions same as in Figure 7.

0.01- 1 5 $

0 I/S 0 2 A3 4 5 6 7 ATP, M x 104 (S) Downloaded by guest on September 26, 2021 VOL. 48, 1962 GENETICS: LEE AND ENGLESBERG 347

leads to an increased inducible level of L-ribulokinase activity.' Mutation in the L-ribulokinase structural gene, besides causing a deficiency in this kinase, leads to an increased or decreased inducible level of both L-arabinose isomerase and L-ribulose-5-P 4-epimerase activities.1' 2 A study of several of these mutants indicates that the L-arabinose isomerase and L-ribulokinase activity, produced as a result of these "secondary" effects, represent a change in amount of wild-type enzymes, as shown by antigenic analysis and by K, values, pH optima, and specificity. The possibility that these dual effects of mutation in the B gene are due to an isomerase-kinase-linked molecule' is not tenable, since (1) the two enzymes are readily separated on a DEAE cellulose column and there is no evidence of inter- action, and (2) L-arabinose isomerase and L-ribulokinase are not co-precipitated by an L-arabinose isomerase anti-isomerase system. Mutation in the f3-galactosidase structural gene (z) in E. coli K12 lowers the activities (presumably the amounts) of ,3-galactoside permease and transacetylase.9 Because mutation in the,3-galactoside permease gene (y), presumably the structural gene for the permease or perhaps the transacetylase, has no effect on the level of ,f-galactosidase, Jacob and Monod9 proposed a polarity of effect in a direction away from the operator. The dual effects of mutations in both the A and B structural genes in L-arabinose metabolism argue against this polarity. Except for some recent work,'5 evidence for other dual effects is based solely upon assays of enzyme activity.0, 11 Although the discovery of regulatory genes has modified the view concerning the nature of gene control over enzyme synthesis from "one gene, one enzyme" to "one structural gene, one enzyme; one regulatory gene, one or more enzymes," implicit in present thinking is the view that one gene still is involved in one function.6 On the basis of our results, it appears that besides repressor genes9' 12 that may have only regulatory function, i.e. are involved only in the synthesis of repressor RNA and no protein, there are other regulatory genes that are intimately concerned with the structure of a particular protein molecule. Heretofore, the tendency has been to avoid this dualism. Summary.-Gene A in Escherichia coli B/r is the structural gene for L-arabinose isomerase, and A mutants have increased amounts of inducible wild-type L-ribulokinase. Gene B is the structural gene for L-ribulokinase, and B mutants have increased or decreased amounts of inducible wild-type L-arabinose isomerase. E. Englesberg would like to thank C. Yanofsky for his advice in the use of DEAE cellulose chromatography and the Society of American Microbiologists for a President's Fellowship under whose tenure this, as well as other techniques, were acquired. * This investigation was supported in part by a research grant from the National Science Foundation and by a contract from the Office of Naval Research to the University of Pittsburgh. Reproduction in whole or in part is permitted for any purpose of the United States Government. Abbreviations: DEAE, diethylaminoethyl; ATP, adenosine triphosphate; DPNH, reduced diphosphopyridine nucleotide; Tris, tris (hydroxymethyl) aminomethane; EDTA, ethylene- diamine tetraaeetic acid. t Predoctoral Research Fellow, U.S. Public Health Service. I Englesberg, E., J. Bacteriol., 81, 996 (1961). 2 Englesberg, E., R. Anderson, R. Weinberg, N. Lee, P. Hoffee, G. Huttenhauer, and H. Boyer, J. Bacteriol. (in press). Gross, J., and E. Englesberg, Virology, 9, 314 (1959). Downloaded by guest on September 26, 2021 348 GENETICS: V. W. WOODWARD PROC. N. A. S.

I Boyer, H., E. Englesberg, and R. Weinberg, Genetics, in press (1962). 6 Jacob, F., D. Perrin, C. Sanchez, and J. Monod, C. R. Acad. Sci. (Paris), 250, 1727 (1960). 6 Jacob, F., and J. Monod, J. Mol. Biol., 3, 318 (1961). I Cohen, M., in Methods of Medical Research, vol. 5, ed. A. C. Corcoran (Chicago: Year Book Publishers, 1952), p. 301. 8 Toward the conclusion of this study, ara-2, an A group mutant, was employed for the production of pure L-ribulokinase. This was made possible with the finding that this mutant did not produce any L-arabinose isomerase cross-reacting material (CRM) but did produce hyper amounts of wild-type L-ribulokinase. The purification procedure was essentially similar to that described for the isolation of L-arabinose isomerase except that the effluent from the DEAE cellulose column containing the L-ribulokinase was collected. 9 Jacob, F., and J. Monod, in Cellular Regulatory Mechanisms, Cold Spring Harbor Symposia on Quantitative Biology, vol. 26 (in press). 10 Kalekar, H. M., K. Kurahashi, and E. Jordon, these PROCEEDINGS, 45, 1776 (1959). 11 Soffer, R. L., J. Bacteriol., 82, 471(1961). 12 Pardee, A. B., F. Jacob, andJ. Monod, J. Mol. Biol., 1, 165 (1959). 13 Recent assays of similar preparations indicate that the enzyme L-ribulose 5-phosphate 4- epimerase was also present. 14 The resolution of the isomerase and kinase on the larger column was not as good as that achieved with the smaller column, and therefore the amount of Kinase-free isomerase recovered was somewhat less. 16 YanofskY, C., D. R. Helinsky, and B. D. Maling, in Cellular Regulatory Mechanisms, Cold Spring Harbor Symposia on Quantitative Biology, vol. 26 (in press).

COMPLEMENTATION AND RECOMBINATION AMONG PYR-3 HETEROALLELES OF NEUROSPORA CRASSA BY VAL W. WOODWARD RICE UNIVERSITY, HOUSTON, TEXAS Communicated by David M. Bonner, January 5, 1962 Heterocaryons between heteroallelic mutants of Neurospora permit of a partial test (trans only) for functional allelism, the test telling only whether a given pair of nuclei have nonidentical functional impairments. The failure of two mutants to complement means only that their mutant lesions, probably at the protein level, overlap, not that they are identical. Additional criteria of nonidentity follow when two mutants that fail to complement each other show different patterns of complementation with other mutants. Since the complementation unit, defined here as that segment of the genome the individual mutants of which fail to complement all others within the segment, is often a function of the failure of mutants to complement, it would seem worth while to determine whether all noncomplementation is due to intrinsic properties of the mutants in question or, in part, to extrinsic properties. That certain pyr-3 heterocaryons form on agar but not in liquid medium and that nonlinked compatibility alleles' affect heterocaryon formation are evidence that noncomplementation is not always a phenotype reflecting functional relatedness of heteroallelic mutants. The present paper2 both compares a detailed genetic map with a complementation map and describes a change of complementation pattern by induction of a specific, closely linked mutation. The genetic map3 was drawn on the basis of prototroph Downloaded by guest on September 26, 2021