696 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

in well with the structure3 of the thymine dimer which is linked at these two posi- tions.

We would like to thank Dr. R. Rahn for helpful comments on the manuscript. 1 Salovey, R., R. G. Shulman, and W. M. Walsh, Jr., J. Chem. Phys., 39, 839 (1963). 2 Shields, H., and W. Gordy, these PROCEEDINGS, 45, 269 (1959). 3Beukers, R., and W. Berends, Biochim. Biophys. Acta, 41, 550 (1960). 4 Beukers, R., J. IjIstra, and W. Berends, Rec. trav. chim., 79, 101 (1960). 5 Bollum, F. J., and R. B. Setlow, Fed. Proc., 21, 374 (1962).

COORDINATE VARIATIONS IN INDUCED SYNTHESES OF ASSOCIATED WITH MUTATIONS IN A STRUCTURAL GENE* BY NANCY LEE AND ELLIS ENGLESBERG DEPARTMENT OF BIOLOGICAL SCIENCES, UNIVERSITY OF PITTSBURGH Communicated by Klaus Hofmann, July 25, 1963 Mutations which produce a dual effect causing both a deficiency in one and a quantitative change in inducible levels of other enzymes were first detected in a study of L-arabinose negative mutants of Escherichia coli B/r.1-7 L-arabinose negative mutants of this organism have been grouped by genetic and functional criteria into five genes-A, B, C, D, and E. Genes A-D are closely linked and are located in sequence D, A, B, C, between the genetic markers threonine (thr) and leucine (leu) (Fig. 1). Gene E which controls the L-arabinose permease is un- linked to this region as determined by transduction.8 Genes A and B are the structural genes for the enzymes L-arabinose and L-ribulokinase, re- spectively.7 Mutants in gene C, although deficient in all three enzymes2-4 and the L-arabinose permease,9 are not similar to known operator (o0), regulatory (i8), or

thr |139 53 |2 13 7 4 79 6 16* 74 7fi 26 43 23 63 71 68 14 62 90 5S 46 25 1 29- ?4 9 27 |12 5 3 21 | leu

HC=O CH20H CH80H CH20H HCOH C=O C=8 L_ribulose C=O HOCH L-arabinose _ HO-C-H L-ribulokinase HOC-H 'i-rhosuhate-_ :0OCH HOCH isomerase ~ ilOC-H HiOCH --4enierase H4-C-O11 CH20H CH20H C-H20P3K2 CH20P03K2 L-arabinose L-ribulose L-ribulose 5-ohosnhate D-xylulose 5-nhosnhate

FIG. 1.-E. coli L-arabinose gene-enzyme complex.1 7 -* mutant whose order is ambiguous. permease mutants. These alleles of the C gene probably prevent the production of a necessary for the induction of the enzymes (including permease) in the L-arabinose pathway.10 Gene D probably represents the structural gene for L-ribulose 5-phosphate 4-epimerase. The dual effect mentioned above has been described for mutants in the A and B genes. Mutants in the B gene, which we will be concerned with, produce either Downloaded by guest on September 28, 2021 VOL. 50, 1963 GENETICS: LEE AND ENGLESBERG 697 an increased or decreased inducible level of L-arabinose isomerase, ranging from a specific activity of 3-300. The mutant sites affecting low and high levels are irregularly distributed along the B gene6 and these differences in isomerase levels have been shown to represent differences in the rate of synthesis of an enzyme indistinguishable from the wild type.' With three B mutants tested, the 4-epi- merase levels were shown to increase or decrease in the same direction as the cor- responding isomerase levels.7 This paper presents evidence, from an analysis of 21 B gene mutants, indicating that genes B, A, and D act as a "genetic unit of coordinate expression."" Further- more, it will be shown that the dual effect is probably the direct result of a change in the DNA code of a structural gene, a code which programs the structure of an enzyme and sets limits to the rate of synthesis of this and other enzymes specified by this genetic unit. Materials and Methods.-E. coli B/r wild-type and L-arabinose nonutilizing mutants that map within the B gene have been previously ordered and described." 3, 4, 6, 7The preparation of sonicated extracts for various enzyme assays, including their partial purification, has also been given.3' 6, 7 A modified assay for L-arabinose isomerase, employed throughout this study, has been de- scribed elsewhere.6 L-ribulokinase assay is modified from a procedure of Horecker et al.'2 The reaction mixture contains: Tris-(hydroxymethyl) aminomethane (Tris, Schwarz), 42 Mmoles; gluta- thione (Calbiochem), 4 urmoles; ethylenediamine tetraacetic acid (EDTA), 2 jUmoles; MgCl2, 20 /Amoles; NaF, 20 ,moles; (Schwartz, ATN), 8 ,umoles; and ribulose 1-C-14, 5 /Amoles; and cell extract containing approximately 0.5-5 units of activity, in a total volume of 0.5 ml. The reaction was carried out at 37°. One-tenth ml samples were withdrawn at 0, 3, and 6 min into 0.8 ml of absolute ethanol at 00. One-tenth ml of 1 M barium acetate was added, and the precipitate formed in 15 min at 00 was collected on membrane filters (Millipore, HA 0.45 mm), washed 6 times with 1-ml portions of cold 80% ethanol, dried, and the radioactivity determined with a Nuclear-Chicago Micromil thin window gas flow Geiger counter. Specific enzyme activity was recorded in terms of jmoles/hr/mg protein. Ribulose 1-C-14 was prepared from L-arabinose 1-C-14 (Calbiochem and Nichem) by the method of Englesberg,3 and was further purified by paper chromatography in water-saturated butanol on Whatman #3 paper. The position of ribulose 1-C-14 was determined by radioautog- raphy, and material was eluted with water. Nonradioactive ribulose used as carrier was prepared in a similar manner, and the final purification in this case was accomplished by cellulose column chromatography with water-saturated butanol.1'3 The fractions containing only ribulose were identified by circular chromatography, pooled, lyophilized, and stored at -20°. L-ribulose 5-phosphate 4-epimerase assays were as previously described.4 L-ribulose 5-phos- phate and phosphoketolase were prepared by Dr. Richard Anderson. Glyceraldehyde 3-phosphate dehydrogenase was a commercial preparation (Nutritional Biochemicals). Isocitric dehydro- genase activity was determined spectrophotometrically14 at 300. Proteins were determined by the method of Lowry et al.'5 Anti-L-ribulokinase rabbit serum was prepared according to the method used for anti-L-arab- inose isomerase serum production3 The antigen used was a purified L-ribulokinase prepared from an A gene mutant (ara-2). The immune serum obtained was repeatedly absorbed with an extract of uninduced cells. The absorbed serum precipitates kinase, and kinase-CRM can be assayed by its ability to compete with active enzyme for antibody-combining sites.7 Quantitative determination of L-ribulokinase CRM was performed on 5 B gene mutants. 'The details for one such assay with ara-26 are given below (Fig. 2). In a series of 12 tubes (1 X ) were placed 0.1 ml of antiserum, 0.1 ml of 0.8% NaCl, and 0.05 ml (0.29 mg protein) of a 1:4 dilution of ara-26 sonicated cell extract. To each tube was then added a different amount of L-ribulo- kinase, contained in 0.3 ml, ranging from 4 to 48 units in increments of 4 units. In another series (2X), the same reactants were employed, except that ara-26 extract was only diluted 1:2. In the control series (BSA), bovine serum albumin (0.4 mg) was substituted for the ara-26 extract. Downloaded by guest on September 28, 2021 698 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

i Extracts were diluted in 10-3 M EDTA and 4 30 2X . 10-3 M glutathione. After 4 hr of incubation 25 _ ARA 26 over ice, the precipitates were removed by G 20 centrifugation, and the supernatants were Z 5 assayed for remaining kinase activity as de- °' lo0 BSA scribed above. The amountof active kinase Ir Z 5 remaining~ / in the supernatant was plotted against increasing amounts of active kinase o0 _:; -_' added, and the CRM content of the mutant 0 4 8 12 16 20 24 28 32 36410 44 489 UNITS L-RIBULOKINASE ADDED extract was represented by the distance be- FIG. 2.-Quantitative determination of I, tween the end points of test and control ribulokinase CRM in B gene mutant ara-26 titrations. Thus, the CRM content of 1X, (see text). ara-26 extract is 12 units (a), and that of 2X, ara-26 extract is 21 units (b). The results were averaged and expressed as units of kinase CRM per mg of protein. Each mutant extract was assayed in this manner. A unit of kinase CRM is defined as the quantity that will protect one unit of active enzyme from precipitation by antiserum at the equivalence point. Results.-To determine whether mutations in the B gene, the structural gene for L-ribulokinase, affect the syntheses of L-arabinose isomerase and L-ribulose 5-phosphate 4-epimerase in a coordinate fashion, extracts of wild-type and 21 B gene mutants grown in a casein hydrolysate arabinose medium were assayed for the above enzymes and isocitric dehydrogenase. The latter was employed as a control for determining the general activity of the extracts. In most cases ex- tracts were prepared in duplicate. L-arabinose isomerase activity of each mutant extract was plotted against its 4-epimerase (Fig. 3) and isocitric dehydrogenase specific activities (Fig. 4). It is apparent that there is a wide spectrum of isomerase and 4-epimerase activities among 3-5 ., the B mutants and that these in- MUTANTS IN THE L-RIBULOKINASE 71 / STRUCTURAL GENE *63,/ creases and decreases in enzyme 3.0 -4-EPIMERASE VS. ISOMERASE levels are proportionate or coordi- ,_ (PARTIALLY PURIFIED EXTRACTS) l /U2. nated. The isocitric dehydrogenase ,279, 2 0712 activity, on the other hand, shows 4% t 152 2/-632little*232 variation. 2.0 -151 g n *3I Five L-arabinose-inducedprdceaoutB gene WT2 WTI mutants produce significant amounts 1.5 of L-ribulokinase CRM. Extracts /*26 of these mutants were assayed for 10 = I~o;L-arabinose isomerase, L-ribulo- 0.5 /43 8" kinase CRM, and isocitric dehy- drogenase, and the specific activities 10 55.6S,342,25) of these enzymes were plotted in 0`2 50 100 150 200 250 300 35o a similar manner as above. The L-ARABINOSE ISOMERASE SPECIFIC ACTIVITY levels of isomerase and of kinase FIG. 3.-Coordinated syntheses of L-ribulose 5- CRM appear coordinated, over the phosphate 4-epimerase and L-arabinose isomerase in wide range of activities exhibited by wild-type and B gene mutants. Numbers are mu- tant designations, and subscripts 1 and 2, when these mutants (Fig. 5). The iso- present, represent different extracts of the same citric dehydrogenase assays again mutant. Assays were performed on extracts pre- pared by the method of Cribbs and Englesberg6 and showed little variation from one partially purified by MnC12 precipitation, resulting mutant to the other (maximum 2- in approximately a twofold purification without any fold). loss of activities. fold). Downloaded by guest on September 28, 2021 VOL. 50, 1963 GENETICS: LEE AND ENGLESBERG 699

The five kinase CRM producers 350 among the B gene mutants were STRUCTURAL GENE selected from a qualitative screeningcen - 300 -ISOCITRIC DEHYDROGENASE VS. ISOMERASE - selected~~~~~~~~frmaqaiatv (PARTIALLY PURIFIED EXTRACTS) test. There are a number of other ( B mutants which seem to have Q 250 traces of CRM as previously re- A ported,7 much below the level of , 200 ara-29. Some of these mutants are high isomerase and 4-epimerase 1 producers (ara-24, ara-71). We , 00 would predict, on the basis of the 74, 46,0462 coordinate control that we have OoTI- 32 C shown, that such mutants are prob- 50 2h 6i..2 2. ..2e2% 71iI.. ably producing an amount of __2_ kinase-like protein proportional to 0 50 100 150 200 250 300 350 L- ARABINOSE ISOMERASE SPECIFIC ACTIVITY their isomerase and 4-epimerase FIG. 4.-Isocitric dehydrogenase specific activ- levels. The extremely low levels of ities of wild-type and B gene mutants plotted against their respective L-arabinose isomerase kinase CRM which were detected levels. Numbers and subscripts are as in Fig. 3. would be due to a loss in both en- zymatic activity and antigenic cross reactivity resulting from the structural rear- rangements in the mutant kinase protein. We have previously shown that the increases and decreases in rates of enzyme synthesis as a result of mutation in the kinase structural gene are not due to the production of altered enzymes.7 Neither is it possible to explain the varied activi- ties on the basis of differences in the ability to take up and concentrate inducer. Novotny and Englesberg,9 using C-14 xylose as a measure of the L-arabinose T0 l I MUTANTS IN THE L-RIBULOKINASE GENE permease,preswrunbetdeosrtwere unable to demonstrate w 60 _KINASE CRM VS. ISOMERASE 23 any great variation in the Larabinose L> (CRUDE EXTRACTS) permease among the B mutants, and where ui 50 _ exists, about 2-fold, 26 a small difference 40 43- it did not parallel the isomerase levels. He / Also, the residual kinase activity, or 0 30 _ leakiness, which might conceivably alter / the intracellular concentration of L-arabi- 2 WT nose, does not appear to be associated ° 10 -29 with either high or low enzyme levels Zu) (Fig. 6). Furthermore, mixing experi- 0 20 40 60 80 100 120 140 ments have apparently ruled out the L-ARABINOSE ISOMERASE SPECIFIC ACTIVITY possibility of the production by these FIG. 5.-Coordinated syntheses of L- mutants of a cytoplasmic inhibitor or arabinose isomerase and L-ribulokinase CRM. activator of enzyme activity.3I Experi- activity Wild-typewas also plottedL-ribulokinaseagainst itsspecificisom- ments with sexual merozygotes have failed erase specific activity for comparison. totodemonstratedmonsratethethe presence of aanyy cytcyto- perThe mgmannerof proteinin whichforthevariouskinaseBCRMgeneunitsmu- plasmic B gene product that affects the tants are obtained is as depicted in Fig. 2. rate of synthesis of L-arabinose isomerase earlierExtractsmethodemployedwhichwereyieldedpreparedcells ofbyrela-an or L-ribulokinase. 10 tively lower activities.4 Downloaded by guest on September 28, 2021 700 GENETICS: LEE AND ENGLESBERG PROC. N. A. S. Discussion. -The coordinate synthesis of L-ribulokinase CRM, L-arabinose isomerase, and L-ribulose 5-phosphate 4-epimerase, as demonstrated with the dual effect mutants in the B gene, indicates that the structural genes concerned (B, A, and D) represent a "genetic unit of coordinate expression."1 Coordinate control of enzyme synthesis was first discovered in a study of histidine biosynthesis by Ames and Garry,"6 and in ,3-galactoside utilization by Jacob and Monod.'1 Frank- lin and Luria,"7 Jacob and Monod,11 and more recently Ames and Hartmanl8 have also demonstrated a dual effect of mutations in structural genes. In the histidine and ,3-galactosidase systems, such mutations only affect a coordinate decrease,

0.50 79

> 0.40

,, 0.30 _

4 0.20 _

0.10 _ 24 TO 25 43 14 29 8 4 1526 C 283 22 i 62 71 0. 56 50 100 150 200 250 300 350 L-ARABINOSE ISOMERASE SPECIFIC ACTIVITY FIG. 6.-Residual L-ribulokinase specific activities of B gene mutants ("leakiness,") plotted against their respective L-arabinose isomerase specific activities. Numbers are mutant designations, and values of their L-ribulokinase specific activities are based on the average of two assays performed on two different extracts of the same mutant.

but never an increase, in level of enzyme synthesis. There has also been no evidence presented so far in these sytems for coordination of CRM formation (self regulation). The absence of mutations in the f3-galactosidase structural gene and the histidine structural genes leading to high inducible levels of enzyme may be due to specific differences of these three systems, and also, in the case of the histidine system, the conditions under which the enzymes were assayed. In the f-galactosidase system, mutants in the z gene with high levels of permease might be inhibited by lactose which would accumulate within these cells in large quantities, thus preventing the recovery of such mutants in the usual mutant isolation procedures. This problem has not been a factor in the isolation of L-arabinose negative, ribulokinaseless mutants producing high levels of isomerase and epimerase, since the L-arabinose permease gene is unlinked to the threonine, arabinose, leucine region of the chro- mosome, and the permease levels are not affected to any great extent by mutation in this structural gene.8 9 In the histidine system, mutants containing high levels of enzyme due to mutation in one structural gene would accumulate large con- centrations of intermediates in histidine biosynthesis which might inhibit growth and prevent their isolation. Whether such mutants are present or not might also be obscured by assays for the enzymes concerned under conditions of repression.'8 Downloaded by guest on September 28, 2021 VOL. 50, 1963 GENETICS: LEE AND ENGLESBERG 701

In normal circumstances the maximum rate of enzyme synthesis may be governed by repressor concentrations. The evidence presented so far precludes the possibility that the dual effect caused by a mutation in one structural gene is due to the production or lack of production of a particular cytoplasmic product (repressor, inhibitor, inducer, or ac- tivator) affecting the phenotypic expression of the other genes concerned. It is also apparent that the results presented are not compatible with a model which pre- dicts that the kinase structural gene, for instance, produces a polypeptide which is shared by both kinase, isomerase, and epimerase. Such a model would not explain the effect of such mutations in causing coordinate changes in rates of synthesis of wild-type enzymes. Also the absence of any immunological cross reactivity of the isomerase and kinase places further doubt on such a model. The dual effect can best be explained, at present, on the basis of a model that predicts that the coding of the structural gene in a "genetic unit of coordinate expression" controls directly the rate of synthesis of an enzyme for which it itself bears structural information as well as for the rate of synthesis of the other enzymes involved. Mutation in structural genes affecting rates of protein synthesis have been previously postulated'9 to explain certain cases of abnormal hemoglobins. The controlled coordinate synthesis of the three enzymes in L-arabinose metab- olism can be achieved by the transcription of a large messenger RNA (mRNA) produced in a polar fashion beginning at one end of the kinase structural gene (toward the leucine marker) and copying in a sequential manner the B, A, and D genes. This completed intact message would then be translated by the soluble RNA-activating enzymes-ribosome complex into the three enzymes. This model is similar to one proposed by Jacob and Monod" for the lactose operon. The DNA code of one structural gene may determine the rate of synthesis of several enzymes by its effect on at least three factors in protein synthesis. (1) The DNA code may affect the efficiency of transcription of this large genetic unit of DNA by the DNA- dependent RNA . (2) The mRNA code in turn may determine the stability of the mRNA. (3) The mRNA code may determine the efficiency of the translation of mR'NA into protein.'8 Since ribosomes have an apparent stabilizing effect on mRNA, factors 2 and 3 are obviously related. Changes in the code due to mutation therefore could result in increases, decreases, or perhaps have no effect at all on the rate of synthesis of the protein involved. Neither of these three parameters involved in protein synthesis can be eliminated a priori as playing a role in the dual effect and obviously they are not mutually exclusive. Since the rate of enzyme synthesis and structure of an enzyme may be thus intimately related, selection may result in a compromise between maximum rate of enzyme synthesis (which may not be optimum for growth) and the properties of the enzyme as a catalyst. It might be possible on the basis of this hypothesis to have alterations in the code of a structural gene which would result in increased or decreased rates of synthesis of an enzyme complex, like the L-arabinose one, with- out affecting the enzymatic activity of the kinase. This model also allows for the synthesis of different amounts of enzymes produced by one large multicistronic message. With the determination that the mutant alleles of the C gene that have so far been characterized do not function as o0 mutants, this gene-enzyme complex of Downloaded by guest on September 28, 2021 702 GENETICS: LEE AND ENGLESBERG PROC. N. A. S.

isomerase, kinase, and epimerase, although acting as a "genetic unit of coordinate expression," is left without a "classical" operator locus, and therefore cannot be considered as an operon sensu stricto. However, since mutant sites such as ara-14, ara-55, ara-25, and ara-27 are operator-like mutants (o0) having very reduced levels of the three enzymes, there may not be a real distinction between the o(Ltype mutants which occur at the beginning of an operon and the other mutants within the B gene which we have mentioned.'8 The findings that the operator locus of the 3-galactosidase operon probably contains structural information for this enzyme" further supports this contention. Summary and Conclusions.-Mutations located at different sites in the L-ribulokinase structural gene cause coordinated increases or decreases in the syn- theses of L-arabinose isomerase, L-ribulose 5-phosphate 4-epimerase, and L-rib- ulokinase CRM. The possibility of this dual function of a structural gene being caused by altered enzymes, altered internal concentration of L-arabinose and cyto- plasmic inhibitors, repressors, activators, or inducers all appear to have been ruled out. It thus appears that the rate of enzyme synthesis is not solely determined by regulatory genes of the i type but the coding of the structural gene may set the limits in the rates of synthesis of one or more enzymes specified in a "genetic unit of coordinate expression." We would like to thank Dr. Richard Anderson for his generous gift of L-ribulose 5-phosphate and phosphoketolase, and Drs. P. Margolin and E. Umbarger for their critical review of the manuscript.

* This investigation was supported in part by a National Science Foundation research grant G11332, USPHS research grant GM10165 from the Division of General Medical Sciences, U.S. Public Health Service, 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. Brief summaries of this work have appeared in Bacteriol. Proc. (1963), p. 35, and in Abstr. Commun., 11th International Congress of Genetics, 1963, The Hague, The Netherlands. 1 Gross, J., and E. Englesberg, Virology, 9, 314 (1959). 2 Englesberg, E., and N. Kileen, Genetics, 44, 508 (1959). 3Englesberg, E., J. Bacteriol., 81, 996 (1961). 4 Englesberg, E., R. L. Anderson, R. Weinberg, N. Lee, P. Hoffee, G. Huttenhauer, and H. Boyer, J. Bacteriol., 84, 137 (1962). 6 Boyer, H., E. Englesberg, and R. Weinberg, Genetics, 47, 417 (1962). 6 Cribbs, R., and E. Englesberg, Genetics, submitted for publication. 7 Lee, N., and E. Englesberg, these PROCEEDINGS, 48, 335 (1962). 8 Isaacson, D., and E. Englesberg, unpublished data. 9 Novotny, C., and E. Englesberg, unpublished data. 10Helling, R. B., and R. Weinberg, Genetics, in press. "1 Jacob, F., and J. Monod, in Cellular Regulatory Mechanisms, Cold Spring Harbor Symposia on Quantitative Biology, vol. 26 (1961), p. 193. 12 Horecker, B. L., J. Thomas, and J. Monod, J. Biol. Chem., 235, 1580 (1960). 13 Englesberg, E., Arch. Biochem. Biophys., 71, 179 (1957). 14 Englesberg, E., these PROCEEDINGS, 45, 1494 (1959). '5Lowry, O., N. Rosebrough, A. Farr, and R. Randall, J. Biol. Chem., 193, 265 (1951). I' Ames, B. N., and B. Garry, these PROCEEDINGS, 45, 1453 (1959). 17 Franklin, N. C., and S. E. Luria, Virology, 15, 299 (1961). 18 Ames, B. N., and P. Hartman, in Synthesis and Structure of Macromolecules, Cold Spring Har- bor Symposia on Quantitative Biology, vol. 28 (1963), in press. 19 Itano, H. A., Advances in Protein Chem., 12, 215 (1957). Downloaded by guest on September 28, 2021