Regulation of the gal Operon of Escherichia coli by the capR Gene*

SUMMARY We have examined in greater detail the derepression of uridine diphosphate galactose-4-epimerase synthesis in capR mutants of Escherichia coli first observed by Markovitz.


GEORGE MACKIE AND DAVID B. WILSON
From the Section of Biochemistry and Molecular Biology, Division of Biological Sciences, Cornell Uwiversity, Ithaca, New York 14850 SUMMARY We have examined in greater detail the derepression of uridine diphosphate galactose-4-epimerase synthesis in capR mutants of Escherichia coli first observed by Markovitz. (MARKOVITZ, A. (1964) Proc. Nat. Acad. Sci. U. S. A. 51,[239][240][241][242][243][244][245][246]. All three enzymes of the gal operon are derepressed from 2-to 4-fold by this mutation. This derepression may be superimposed on the conventional derepression seen when the gal operon is induced with fucose or by a regulatory mutation. Measurement of the level of gQl messenger ribonucleic acid indicates that the derepression caused by the capR mutation probably occurs at the level of transcription since there is a coordinate increase in gQl enzyme activity and in the level of gal mRNA. The effect of the capR mutation on the gal operon appears to be independent of cyclic 3',5'monophosphate or glucose-mediated repression. Thus the product of the capR locus behaves as if it is an additional negative control element decting the expression of the gal operon.
Mutat,ions in either galR or gal0 result in the constitutive synthesis of all t,hree enzymes of the operon.
There is no firm evidence for a promoter region, but several mutants have been described with properties similar to those expected of strains carrying promoter mutations (5, 6). It has also been proposed that a positive mechanism of control involving cyclic adenosine 3', 5'monophosphate and the cyclic AMP1 receptor protein (7) also acts on the gal operon, presumably at' a promoter site (8-10).
'I'hc others include phoaphomaunoseisomerasc, Gl )l' mauuose hydrolyasr, (;I )I'-L-fucose synthetase, GDP mannosc I)yroI)hosl)hor~l:Lse, TTl)l' glucose dehydrogenase, and 1DP glucose l,yrol)hosl)horylase (11,17). Markovitz has postulated that the product of the cnpR locus is a repressor protein which binds to the operator region of the structural genes for each of t.he above enzymes. Loss of the product of the cupR locus results in the derepression of all of the genes normally regulated by this protein.
We undertook to characterize further Dhe effect of t,he product of the cupR locus on t,he gal operon.
First, we determined whether or not the ent,ire operon responded to the product of the cupR gene and, if so, whether this response occurred at the level of transcription or translation. In fact, all three enzymes are derepressed in cupR mutants, with a corresponding increase in the level of gal mR?iA.
Second, we demonstrated that the effect of mutations in cupR is for the most part independent of other regulatory phenomena in the gal operon such as induction and glucose-mediated repression. Finally, we have attempted to distinguish between the model of negative control suggested by Markovitz and other models involving positive mechanisms.
We conclude that Markovitz's original proposal is consistent with our results.

MATEKIALS 4x1) METHODS
Strains-Bacterial strains used in this work are described in Table I. Bacteriophage Xels~i was obtained from lysat.es of W3350 (&S;) and was used to construct Xdg30 which retains the cl857 marker and which transduces the entire gal operon. Stocks of Xpgu& were obtained from Dr. Ira Pastan (National Institutes of Health).
This phage carries a deletion of the host chromosome between t.he bact,erial attachment site (attB> and the gal operon (18). Growth Media-H-l salts consist of 11.2 g of K2HPOI, 4.8 g of KH2P0+ 2.0 g of (NH&S04, 1.0 ml of 1.8 mM FeS04, and water to 1 liter. MgClz to 1 mM, carbon source, and supplements were added after autoclaving.
Mannitol was most often used as a carbon source, as it allowed good growth for the strains under study and did not interfere with inducer entry. Tryptone broth consists of 10 g of Difco Bacto-tryptone and 5 g of NaCl per liter.

2974
Regulatim of the gal Opelo?r Vol. 247, No. 10 Strains GM91 and GM100 were isolated from st,rains 3300 and M6, respectively, by IRCl21 mutagenesis followed by two cycles of growth in thiomethyl galactoside galactose minimal media. Strains GM130, GM131, and GM132 were const.ructed by transducing a gal-derivat.ive of 3300 to gal+ using Pi phage grown 'on the indicated galOc strains.
Strains GM140, GM141, and GM142 were constructed as above, except a gal-derivative of M6 was used. Growth of Bacteria---Hacterial strains were maintained on tryptone broth stabs. For measurements of enzyme activit,y, fresh stationary broth cultures were dilut.ed 1 :lOO into H-l medium, grown as indicated in each table, and harvested while St.11 ,u exponential phase.
Growth of Bacteriophage-Btrains W3350 (X,lsa?) and GM30 were grown in tryptone broth at 28" in a fermentor or in 4-liter flasks with vigorous aeration. When t.he cells had grown to a density of 5 x lo* cells per ml, they were induced by heating for 15 min to 42" and then cooled to 37" where growth was continued until lysis occurred.
Debris from the lysate was removed by centrifugation in a Sharples refrigerated centrifuge. Phage were recovered from the supernatant by prolonged centrifugation (16 hours) in a Sorvall GS-3 rotor at 8500 rpm. The phage pellet was suspended in X dilution buffer (19) and subjected to a cycle of low and high speed centrifugation.
The final phage suspension was then purified by centrifugation to equilibrium in a cesium chloride gradient (19). Xpga& was grown by infecting a stationary phase culture of strain N99 in tryptone broth containing 0.2Cz maltose and 1 mM MgSOa with phage at a multiplicity of 5 to 10. After adsorption, the mixture was diluted 1:30 into fresh warm tryptone broth (3i"), and growth continued for several hours (S mutants are unable to lyse their hosts). The phage-infected cells were collected by cpntrifugntion and lysed with chloroform, and the phage was purified from the concentrated lysate by the method described above. (Th' 1s ec n1 t h 'q ue was suggested to us by Dr. Pa&n, National Institutes of Health.) Isolation of DNA---I)NA was isolated from purified bacteriophage by t,he method of Kaiser and Hogness using buffersaturated, freshly distilled phenol (19).
To this were added (for 3 to 5 x log cells) 200 pg of egg white lysozyme and 25 pg of deosyriboli~lcleaxe. This mixture was rapidly frozen and thawed three times, then acidified with 1.0 ml of 0.5 M sodium acetate, pH 5.2. RIacaloid (21) and sodium dodecyl sulfate were added to final C'OIKPI~trations of 0.2 and 0.1%) respectively.
After 5 mill at room temperature, the mixture was extracted at 60" w&h vigorous mixing with an equal volume of hot phenol satulatcd wit.h 0.2 M sodium acetate, pI-I 5.2. The aqueous phase was rccovned from the emulsion by centrifugation and extracted twice mow. lsefore each of these cstractions 2 drops of dietl~~l~~yroc~al~l~o~lnte were added as a nuclease iuhibitJor (22). 'l'hc final aqueous I)hnse was dialyzed extensively against sterile 2 x SSC to wnmw the phenol.
DNA-RNA Hybridization-111 general, we followed t.he procedures of Spiegelman and Gillespie (24) wi6h the followiIlg modifications.
DNA was denatured with alkali, neutralized, and loaded onto 13.mm uitroccllulose filters (type n-6, Carl Schleicher and Schuell Company) at 5 or 10 pg per filter. Retention of DNA on the filter was usually over 95';;. DNA was fixed to t,he filters by drying them under vacuum, first for 1 hour at room temperature and then for at least 2 hours a.t 65".
Components of the hybridization mixture included [%-RNA, usually from 2 to 5 pg depeuding 011 the specific activity; 30 pg of unlabeled ribosomal RNA to act as carrier and to compete against nonspecific binding; and buffer (2 x SSC, 25. neut,ralized phenol, 0.05Yi;;. sodium dodecyl sulfate, pH 6.9) to 0.25 ml in a 2-dram vial. d filter containing the alqxopriate DNA was wetted with a minimal volume of buffer and added last. The vial was sealed carefully and incubated for 24 hours at 65". At the end of this period, the filter was removed from the vial and washed with 2 X SSC (15 ml per side). The filter was incubated for 40 to 50 miu at 32" wit,h 20 pg of ribonuclease in 1.0 ml of 2 x SSC and t,hen rewashed.
The filter was dried thoroughly before counting iu a toluene-based liquid scintillation fluid. The rcaults arc expressed as t,he percentage of input counts hybridized.
In any one esperiment,, all filters were from the same lot.
One unit of activity equals 1 pmole of product formed per hour except for /3-gnlactosidase where 1 unit equals 1 nmole of 0-nitrophenol per min. Extracts for these assays were prepared by suspending each frozen cell pellet (frozen at -20") in 1.5 to 2.0 ml of a buffer containing 0.02 M triethanolamine acetate, pH 7.9, 1 m&f EDTA, and 1 mM dithiothreitol.
The suspension was transferred to a. polyallomer centrifuge tube (1.5 X 7 cm), covered with parafilm, and placed in the chamber of a Raytheon sonicator (DFlOl) containing 30 ml of Hz0 at 4". The sonicator was operated at full power for I nLn, chilled, and operated again for 1 mill more. Synthesis-Markovitz's original work (11) indicated that cpimcrase activity was 4-to s-fold higher in a mucoid (capR-) strain than in the I)arcnt (3300). The data in Table II demonstrate that all threa cnzymcs of the galactose operon are synthesized at a higher rate in the capR st.rain. Growing the cells in a different medium, such as 119 (29), or using a diffcrcnt carbon source, such as glucose, does not substantially alter the results. The epimcrasc appeared to be synthesized at a higher rate than the ot.her two enzymes. However, the higher epimerase Specific activity of epimerase may be an artifact, resulting from the error caused by substant.ial blanks present in the assays for kinase and transfcrnse which partially obscure the basal level of activity in 3300.
Mixing experiments (data not shown) were performed to ascertain whether one of the extracts contains an inhibitor or activator not present in the other. For all three enzymes, activity was completely additive in mixed cstracts. These experiments rule out the presence of a dissociable inhibitor or activator in extracts of these st.rains. E$ect of Interruption of Normal Regulation-It was of interest to determine whether or not the capR-strain used would show higher enzyme levels in situations where the gal rnzymes are normally derepressed. Such situations arise when t.he growth medium contains n-fucose, a gratuitous inducer of the gal operon (a), or when the cells in question carry mutations in the galR or gal0 regions.
The results in Table III indicate that strains carrying n capR mutjntion always l~ossess llighrr levels of cpimerase than do the corresponding parcnt,s. Assays of kinase activity (and in the cast of ~~11s grown with fucose, transfcrasc: activity) gave results directly p:~rallel with those shown.
Transcription in capRand Parent Strain--Rc mawured the synthesis of gal mRNA in both Strain 3300 and St,rnin 316 by hybridization techniques using pulse-labeled K,NS. This method is valid provided that the kinetics of RN,4 synthesis is identical in the strains under study. Experiments demonstrating this for M6 and 3300 are presented in Fig. 1  that the uptake of uridinc from the medium, under conditions similar to those ustd in ljulse-labeling, and its incorporation into acid-insoluble material are nearly identical in the two strains. Table IV shows the results of hybridizing p&c-labeled RNA from strains M6 and 3300 (with or without n-fucose) to different DNAs. Table V shows the relative epimerase levels of these same cultures.
Xthough there is an increase in hybridizable RNA in M6 relative t,o 3300, this increase is much less than the increase in gal enzyme activity. lloreover, although fucose induces the gal enzymes lo-to 13.fold, the increase in gnE mRNA synthesis is only 2$-fold l&en XdgYO DNA is used in the assay and 4-fold when Xpga/8 DNA is used. Xpgal8 carries the entire gal operon, but it' lacks a segment of E. coli chromosomal material carried in Xdg30. This strongly suggests that other RSA species, in addition to gal mRNh, are hybridizing to both types of 1)NA and obscuring the increase in gal mRSA.

Analysis of Hybridixalion
Data-We have analyzed our data making the assumption t,hat, upon induction with fucosc, gal mRNA and gal enzyme levels are directly proportional. For this discussion, we define the following quantities using the enzyme ratios measured in fucose induced cells (Table V) z = y0 hybridizable RNA from 3300 that is not gal mRNA Z' = y0 hybridizable RNA from M6 that is not gal mRNA y = y0 hybridizable RNA from 3300 that is gal mRNA 13~ = y0 hybridizable 1iNA from 3300 grown with fucose that is gal mRNA z = y0 hybridizable RNA from M6 that is gal ml{NA 52 = '% hybridizable RNA from M6 grown with fucose that is gal mRNA Applying these symbols to the da& I'rom Table IV for  These results are striking confirmation of our oripillal Xsumptions.
First, one would : expect z = 2', since these rcpre~ent. the mRNA which is not gal mRN=1 and which should not. respond to fucose. This is the mRNA from tile genes which lie between the phage attachment site on t,he host chromosome (at.&) and the gal operon.
The amount of this message ought to be the same in both strains, and our calculations show it is.
Second, the derepression of gal mRNA in strain ?\I6 relative to 3300 is now virtually identical with the extent of derepression of epimerase which is the best measure of all the gal enzymes (Table V). These data indicate that) the product of thr capR gene affects transcription, that is, it normally rerepresses transcription some 4-fold.
The dat'a furbher imply t#hat if there is translational control in the gal operon it is not. nffert.ed by the capR product.
Third, if the results obtained from using Xpgnl l>Kh in the hybridization assay are substitut,cd int,o t,he equ;ltions developed previously, we obtain the I'ollowing solutions. That is, identical results arc derived for the percentage of go1 mRNA in the mutant and parent strains using different l)S,\s in the hybridization assay. In the case of XpgnB DSA, only the values of z and 2' are reduced as expected since this phage is derived from a strain carrying a deletion bet,ween the phage attachment site and the gal operon (18). It is int,eresting to note that even with t,his deletion, the amount of RNA other than gal mRNA which hybridizes to this DNA is &ill twice as great as the amount of gal mRNh in uninduced cells. Thus it would appear that Xpga/8 phage carries significant amounts of DNA derived from the host chromosome in addition to the gal operon.
Finally, our values for both the basal and induced levels of gal mRNA in strain 3300 are very close to those obtained by Miller et al. (30).  Cells were grown at 29" in H-l salts with thiamine at 10 pg per ml and 0.4% D-mannitol as carbon source. At a density of 1 X 10s cells per ml, the cultures were each divided into two lots, one of which was further supplemented with cyclic AMP to 5 mM. Growth was continued until the cells were harvested at 3 X lo8 cells per ml. The cell pellets were stored at -20" until assayed as described previously. (SSOO) strains Cells were grown to stationary phase in H-l salts containing 10 pg per ml of thiamine and 0.4cr glucose. These cultures were used to inoculate a medium composed of H-l salts (without ammonium sulfate), 10 pg per ml of thiamine, 1 mM MgSOs, 0.597, glutamic acid, 6% tryptone broth, and 0.2% glucose. These cultures were aerated at 30" to a density of 1.5 X 10s cells per ml. One portion of the culture was retained as a control, and the other was supplemented with cyclic AMP to 5 mM. Growth was continued to 4 X lOa cells per ml when the cells were chilled, harvested, and assayed as before. E$eet of Cyclic AM&-There is ample evidence to suggest that cyclic AMP is a positive effector in the process of transcription (7,31,32), and that the lowering of the intracellular concentration of cyclic AMP is the likely cause of the repression of inducible enzyme synthesis during catabolite repression (31,(33)(34)(35).
It is conceivable that concentrations of cyclic AMP might be increased by the capR mutation in strain M6, thereby giving rise to the derepressed rate of transcription of the gal operon in MB.
The data of Table VI indicate that cyclic AMP added to the growth medium does not mimic the effect of a mutation in the capR gene on the synthesis of the gal enzymes. Rather, the specific activky of epimerase in st.rain 3300 is unaltered by growth for 1: generations in the presence of 5 mM cyclic AbIP.
Surprisingly, in M6 the specific activity of epimerase reproducibly falls roughly 30yp under the same growth conditions.
In cont,rast, in both strains cyclic AMP stimulates fl-galactosidase synthesis by 207;. -1s a further test of the relation between positively regulated phenomena and mucoidy, epimerase and ,&galactosidase were assayed in extracts of capR-(hI6) and parent (3300) strains grown under conditions of (permanent) catabolic repression.
In this case, a medium containing a poor nitrogen source was used to accentuat.e the repression caused by glucose (35). A comparison of the data of Table VII with those of Table VI indicat.es that whereas the epimerase in both strains is not subject to catabolic repression under the conditions of the experiment, i.e. at 30", fi-galactosidase is strongly repressed at this temperature.
(It is apparent that cyclic AMP does not reverse the repression of P-galactosidase synthesis in M6 to the same extent as in 3300. The decreased response of /3-galactosidase to cyclic AMP in 316 suggests that M6 may be less permeable to cyclic ARIP than is its parent.) Thus, under conditions of catabolic repression at 30", there is no repression of the gal operon in the parent strain nor any reversal of the derepression of the gcrl operon caused by the capR mutation.

1)ISCUSSION
The model presented by Markovitz (11,12) is collsistcnt with the data presented in this paper. The dcrcprt4on ot' the gal enzymes and of g& mRNh caused by a mutation in t,hc capR gene in strain M6 appears to bc similar to that caused by a mutation in t,he gall? gene. Since we hove not mcnsured the kinetics of gal mRNA synthesis, only its IwcI, we cannot say whether the capR product effects the s;\-nthesia or the degradation of gal mRNA; however, it, seems probable tht\t it is increasing the rate of synthesis by analogy with other regulating systems. The derepression caused by the presence of the capR6 allele occurs even in induced cells, showing t.hat the capR product acts independently of t,he galR product. The magnitude of the derepression of t'he gal enzymes by the cnpR allele is greater in uninduced cells than in induced cells (Table  III).
However the level of gal enzymes and gal mRN.\ in induced capR-cells is very close to the sum of the levels in induced capR+ and uninduced capR-cells. This is the expected result if the control by the capR gene was completely iudependent of the control by the gaZR gene. Another explanation is that E. coli contains two sets of gal genes, the normal set in the gal operon and another set under the control of t,he capR locus. This possibility has been eliminat.ed by the isolation of a transferase-negative derivative of strain 316. This strain la&a transferase when it is grown under conditions in whirh t.he mucoid phenotype is expressed as well as when it is induced lor gcr2 enzyme synthesis.
Furthermore, t'r:bllsdll('t:lllt,~ are re:tdily isolated which have transferase act,ivity under both contlitiolls, indicating that a mutation ill a single locus is responsible for the loss of both the mucoid-and the fuco+illtlucrtl tr:msfer;lse activities in the mutant strain.
Markovitz's proposal that a single repressor-like molt'culc, the product of the capR gent, is capable of regulating the nctivity of a number ol diverse genes is not without precedent.
It simply demands that all of the genes so regulated share a common regulatory site, analogous to an operat.or. The enzymes of arginine biosynthesis (36, 37) exemplify such a cast. What is novel in this model is that the gal opcron must respond to two different repressors, the products of the galR and ca.pR gelles;. Consequently, the gal operon must possess two regions capable of recognizing regulatory proteins, the well known gal0 region and another site which we will designate gal&. It is this latter region which all genes under the regulation of the ca.pR gmc would share.
Evidence from investigations in viva suggests t)hat, the ga,l operon is less sensitive to the presencr of cyclic ;\hlI' than is