Expression of malT,the regulator gene of the maltose regulon in Escherichia coli, is limited both at transcription and translation.

Christine Chapon

UnitedeGenetiqueMoleulaire,InstitutPasteur,25,rueduDr.Roux,75724 ParisCedex 15,France

Six mutations,which lead to an increase in malT expression, were mapped by sequencing techniques. All of them had one or other of two base changes.Determination of the transcription start point by reverse transcriptase mapping localised the two base changes with respect to the elements that control malT expression.One of the base changes(malTpl) is located in the Pribnow box of the promoter,and presumably results in an increase in the rate of transcription initiation. The other (malTp7) is located in the Shine and Dalgarno sequence, which precedes the malT cistron.It probably created a more favourable ribosome binding site on malT mRNA. A correlate of these observations is that the promoter and the ribosome binding site are both in efficient in a wild-type malT gene. A malTpl malTp7 double mutant was constructed, which produced ~30 times more MaIT protein than the wild- type strain.

Introduction
The maltose regulon of Escherichia coli consists of three operons- malPQ, malEFG, and malK lamB- controlled by a positive regulator gene, malT (Debarbouille and Schwartz, 1979; Raibaud et al., 1979; Silhavy et al., 1979; Raibaud and Schwartz, 1980). Experiments with malT-lacZ fusion strains show that the expression of malT itself is positively controlled by 3'-5'cyclicAMP (cAMP) and its receptor, the catabolite activator protein (CAP) (Debar- bouile and Schwartz, 1979). Mutants were isolated, which expressed malT at a high level, in the absence as well as the presence of CAP and cAMP (Chapon, 1982).The mutations map close to the beginning of the malT cistron. Sequencing data presented here show that some of these mutations are located in the promoter of the malT gene,while others are in a region presumed to constitute the ribosome binding site. By combining mutations of the two clases,a strain was constructed that produces ~30 times more MalT protein than the wild-type strain.

Results
Mapping of the mutations by DNA sequencing Mutations causing increased synthesis of MalT protein, in the absence as well as in the presence of CAP, have been designated malTpJ to malTp9 (Chapon, 1982). Their effect on malT expression was deduced from that on (3- galactosidase synthesis when they were located cis to a malT- lacZ hybrid gene. Six of them were studied further. Two classes were found (se also Table I). In strains carying matTpl, matTp8, or malTp9, the level of(-galactosidase was nearly the same in the presence and absence of CAP,i.e.,in a crp+ and crp background. Conversely,in strains carrying malTp4, malTp6, or malTp7, the level of (-galactosidase, even though it was already high in a crp background, was much higher in a crp+ background. The six mutations were transferred to plasmid pOM1, which is derived from pBR322 and carries the malT gene (Figure1)(RaibaudandSchwartz, 1980). It was known (Chapon, 1982) that the mutations mapped between the right end point of deletion malA-511 and that of malAr-510(Figure2). In addition, because of their effect on malT expression, the mutations were expected to map outside the malT coding region,i.e.,to the left of the initiation codon. In other words, the mutations were presumably located in the first 170 nucleotides at the right end of the HpaII-HpaII fragment shown in Figure2. This fragment was purified from plasmids carrying the different mutations,and labelled at both ends using polynucleotide kinase and [-y-32P]ATP. Secondary cuts were made with HincII, and the largest of the two HpaII-HincII subfragments(Figure2) was sequenced by the technique of Maxam and Gilbert (1980). The results are shown in Figure 3. The mutations leading to an almost complete insensitivity to the presence of CAP (malTpl, malTp8,and malTp9) correspond to a GC toTA change 72 base pairs upstream from the initiation codon. The other mutations (malTp4, malTp6, and malTp7) correspond to an AT to GC change, 12 base pairs upstream from the initiation codon. From their position in the sequence, the former groups, including malTpl, could be expected to increase transcription initiation, and the later, including malTp7, to increase translation initiation.

Fig.1. Transfer of the malTp mutations on to plasmid pOMI.The plasmid in (a) is a derivative of pOMI (Raibaudand Schwartz,1980), which caries deletion maMASO (Chapon, 1982).The deletion is represented by a filed box.The chromosomal malT and malP genes are shown below the plasmid, with a malTp mutation represented as a dot. The cel shown in (a) is Tet5 because deletion malAA50 removed the malP promoter, which normally controls the tet genes in pOMI. The recombination events shown in (a) yield TetR cels,which can be selected, and which bear the malTp mutationon the plasmid(b).

Fig.2. A DNA segment that contains the malT and malPQ promoters. The HpaII-HpaII segment shown as a heavy line was previously sequenced (Debarbouilleetal.,1982).The position of a few relevant restriction sites is shown. The numbers are their distance, in base pairs, from the right end of the segment. Deletions malAAS5f and matAS10 are shown as hatched bars below the map. The dashed arrows correspond to the beginning of malT mRNA as determined in this work(seFigure4).

Fig.3. Location of the malTp mutations.The sequence of the right end of the HpaII-HpaII fragment (see Figure2). The base changes corresponding to malT pl and malTp7 are indicated.The interrupted arrow indicates the beginning of the mRNA (see Figure4).Heavy lines indicate the position of the so-called -10(Pribnowbox) and -35 regions of the promoter. The consensus sequences (Rosenbergand Court,1979) for these regions are at the top of the figure. Brackets indicate potential CAP-binding sites, and the interrupted line the Shine and Dalgarno sequence. The numbers under the sequence correspond to the distance expressed in base pairs from the right end of the sequence.

Table 1. Gene fusions are designated 4(matT-lacZ)542-1 (Hyb). The notation [J)(mafT-lacZ)542-1 (Hyb) malTpl] means that the malTpl mutation is located cis to the maIT-lacZ hybrid gene. All strains carrying a gene fusion contain also a X prophage adjacent to the fusion. Pop 2150, pop 2117, and pop2140 were constructed by Raibaud. Strain pop3 is the same as strain MC4100 described by Casadaban (1976). The other strains have been previously described(Chapon, 1982)or constructed in this work.

Fig.4. Reverse transcriptase extension mapping.The hybridization was performed with one of the two 5'end-labelled DNA probes,the 76-hpHpuII-HhaI (partA) or the 19-bp Hpall-Hinfi(partB).50 ug of RNA was used, extracted from strain pop 2117, which harbours plasmid pOM1, carrying the whole malT gene but deleted for the malP promoter, or from strain pop 2140, which harbours plasmid pOM1, carrying the malAA51O deletion (see TableI). The hybridization was done at 300 C for 3.5 h in the presence of 40% formamide (lane 1: pop 2140 (malT), lane2: pop 2117 (malT)], or 80% formamide [lane3:i pop 2117(malT), lane4:pop 2140(maIT-)]. The band indicated by a thick arrow corresponds to the DNA probe extended until the start point of transcription of the malT gene. The interrupted arrow shows the position of the secondary band discussed in the text. Lane5 in (A) shows the G sequencing reaction effected on the 206-bp HpuII-SauIIIA fragment(see Figure3); twice as much of the same material was loaded in lane6 (B). The two DNA probes were also loaded, the 76-bp Hpall1-Hhal fragment in lane0 (A), and the 19-bp HpaxI1-Hinfi fragment which ran out of the gel in lane5 (B). The positions of the 76-bp HpaII-Hhal and the Hpall-SauIlIA fragments are indicated on the figure.The samples were electrophoresed on thin DNA sequencing gel(0.3mm) containing 8% acrylamide.

Table II. The specific activities of (1-galactosidase (units/mg of protein) and (3-galactoside permease (nmols of [14C]thiomethyl ,B-galactoside transported in 20 min by 108cels) were determined in strains where the o(malT-IacZ)542-1 (Hyb) hybrid gene had been introduced in a cis position to the malTp mutation (see Table I). Cells were grown in minimal glucose medium(Glu), in the same medium with 5mM cAMP(Glu+cAMP) or in minimal glycerol medium(Gly).The specific activity of, B-galactosidase in the Acrp strain(see Table I) was the same when cAMP was presentin the growth medium. Numbers in the last column are ratios of the numbers in columns 4 and 5 and not the ratios of numbers of polypeptide chains.

Fig.5. Construction of a malTpl malTp7 double mutant. The techniques used were esentialy as described by Debarbouille and Schwartz (1979) and Hall et al. (1982). The genotypes of the different strains cited are listed in TableI. (a) Transfer of malTp1 ontoi a X phage which carries a malT-lacZ hybrid gene. A malTp1 strain(pop 3931) was firstly lysogenized with a xpo(maIT-lacZ)542-1 (Hyb) transducing phage which contained the amber mutation malT 250(indicated as a cros). ( Debarbouille and Schwartz, 1979). The structure of the resulting lysogen (pop 3951) is shown in (a). The malTpl mutation is represented by a dot. Excision, as shown by the thickarrows, yields the required X phage,which gives dark blue plaques when plated on a Lac-indicator strain (pop3) in the presence of the indicator5-bromo4-chloro-3-indolyl-,3-D-galactoside(X-gal). (b) Lysogenization of a malTp7 strain with the X maITp7 phage. The phage constructed as described above was used to infect a malTp7 strain deleted for the lac operon (pop 3937). The malTp7 mutation is indicated as a dark square. There combination event shown in (b) yieldslysogens that express the hybridgene at the normal low level, and which are therefore pale blue on X-gal containing medium. In lysogens resulting from other recombination events the hybrid gene is controlled by a promoter region containing malTpl or malTp7 and they are dark blue on the same medium. Some of the pale blue colonies were shown to synthesize a mylomaltase constitutively at a high level. They were assumed to produce large amounts of MalT protein and to have the structure shown in (b), third line. (c) Curing of the prophage. Spontaneous Lac-segregants of the above strain were obtained. All had lost X immunity and -30% of them stil expressed a mylomaltase constitutively. These were presumed to result from the excision event shown by thick arrows in (c) and to carry the malTpl and malTp7 mutations cis to a wild-type malT gene.

Fig.6. Detection of the MalT protein in the double mutant strain malTpl malTp7. Cells were grown in synthetic medium containing glycerol, harvested at an optical density at 600nm of 1.0 and concentrated 20-fold in 2x 10-2M Tris HCIpH7.5,10-2M MgCl2,10-4MEDTA,5 x 10-2M NaCl,2.8x 10-3M ,B-mercaptoethanol. A portion of the cels was directly boiled in sample buffer O anef: pop 3(malT+),lanec: pop 3971(maITplmalTp7).The remaining cells were sonicated and centrifuged for 1 h at 20000 g. The resulting pellets were resuspended in the same volume as that of the supernatant in the buffer described above. The supernatant (lanei c: pop 3, lane d: pop 3971) and the pelet (lane a:pop 3, lane b: pop 3971) were boiled in sample buffer. The samples were analysed on 10% polyacrylamide gels in the presence of SDS as previously described (Debarbouille et al., 1978). The position of the MalT protein (ebarbouille et al.,1982) is indicated by an arrow.

Fig.7. Potential ribosome binding sites on malT and malTp7 mRNAs. Complementarities between the 3'-terminal sequence of the 16S rRNA and the mRNA sequence of a malT+ or malTp7 mRNA are shown.The base substitution (A-G) resulting from mutation malTp7 is indicated on the mRNA sequence by an asterisk.The 3'-terminal sequence of 16S rRNA is shown above the malT' mRNA or below the malTp7 mRNA and the base pairing posibilities are indicated.The N-terminalamino-acid sequence of the MalT protein is shown between the mRNA sequences(Debarbouilleet al.,1982).

Identification of the transcription start of the malT gene on the nucleotide sequence
The promoter for gene malT could not be identified by mere examination of the nucleotide sequence. The fact that strains bearing deletion maIAASJJ (see Figure2) still expressed malT,simply indicated that the promoter was to the right of this deletion (Debarbouille et al.,1982). The promoter was precisely mapped by localizing on the sequence the origin of transcription. A specific DNA probe was elongated with reverse transcriptase using total in vivo mRNA as a template (Sollner-Webb and Reeder, 1979; Debarbouille and Raibaud, in preparation). Total RNA was extracted from cel sharbouring a multicopy plasmid which carried gene malT, or cels harbouring the same plasmid except that the malT promoter was deleted. The malT-specific mRNA was then allowed to hybridize with a 5' end-labelled DNA probe (the 19-nucleotideHpaII-HinJlor the76-nucleotideHpaII-HhaI fragments,see Figure2),andthehybridizedDNA probewas extended using reverse transcriptase. The start point of transcription was determined by comparing the length of the extended probe with that of the fragments obtained after chemical cleavage using the Maxam and Gilbert technique.
The results with the two DNA probes clearly indicated that cels harbouring malT, on a multicopy plasmid, contained mRNA molecules with 5' sequences complementary to a DNA segment corresponding to the 89-terminal nucleotides on the HpaII fragment (Figure 4). (Very low amounts of the same mRNA were detected when the plasmid carried a deletion of the malT promoter: these residual mRNA molecules were presumably transcribed off the chromosomal copy of the malT gene, which was intact).A minor population of mRNA allowed the DNA probes to be extended by 86 instead of 89 terminal nucleotides as shown by the band of weak intensition the gel (Figure4). These slightly shorter mRNAs might result from initiation at a secondary origin of transcription of malT, but more probably they result either from a specific degradation of malT mRNA or staggering by the reverse transcriptase. 

From this experiment we concluded that malTpl is located 12 nucleotides upstream from the transcription start,i.e.,in the Pribnow box (Rosenberg and Court, 1979) of the malT promoter,while malTp7 is located in a transcribed region and may therefore affect translation initiation.
A malTp1 malTp7 double mutant produces 30 times more MalT protein than wild-type 
If the effects of malTpl and malTp7 are exerted at different levels of gene expression,these two mutations should have acumulative efect when combined to getherincis. The double mutant was constructed (Figure 5) and a malT-lacZ hybrid gene was introduced cis to the mutations, as previously described (Debarbouille and Schwartz, 1979; Chapon, 1982).The resulting strain produced -30 times more(- galactosidase than the control strain, which contained the samehybrid gene, but no malTp mutations (TableI).
The amount of 3-galactosidase found in the strain that contained the two mutations cis to a malT-lacZ hybrid gene was ~25% of that found in a fully induced lac+ strain (Silverstone et al.,1969).This suggested that the amount of MalT protein synthesized by a malTpl malTp7, but otherwise malT+ strain,might be suficient for the protein to be detectable by gel electrophoresis of crude extracts. This is indeed the case (Figure6) while the protein was not detectable in extracts of the single mutant (Chapon,1982,and unpublished data). The MalT protein in extracts of cells that bore gene malT on a multicopy plasmid had been found to be insoluble in the absence of detergent (RaibaudandSchwartz,1980).

The same was true with a malTpJ malTp7 double mutant. Most of the MalT protein sedimented in 1 h at 20 000 g. All strains previously shown to produce increased amounts of MalT protein expressed the maltose operons constitutively, at levels that depended upon the actual concentration of MaWT protein in the cel (Debarbouille and Schwartz, 1980). Strains bearing malTpl or malTp7 synthesized amylomaltase (malQ product) constitutively only at -25% of the fully induced level (Chapon, 1982). In the double mutant, on the other hand, amylomaltase synthesis was fully constitutive (not shown). This result, which was predictable, was used in a plate test during the construction of the double mutant (see Figure5).

Effect of malTp1 and malTp7 on lacY expression in malT-ñac hybrid operons
A mutation affecting transcription and a mutation affecting translation might be expected to have different effects in the expression of a polycistronic operon. This could be tested in stains containing malTpl or malTp7 cis to a malT-lacZ hybrid gene, because these strains contained lacY, the gene coding for ,B-galactoside permease, as part of the same operon as the hybrid gene (Table I). Mutation malTpl, whichisbelievedtoincreasetherateoftranscriptioninitition, increased lacY expression by approximately the same factor as matT-lacZ expression. However, malTp7, which is assumed to increase the rate of translation initiation, had much less effect on lacY expression than on B-galactosidase synthesis.

Discussion
The six mutations we studied, which increase malT expression, were obtained independently. However, they all correspond to one or other of two base changes. This result may indicate that only few base changes, and perhaps only these two,can lead to a significant increase in malT expression.
Mutation malTp1 and the other mutations of the same base change increase the rate of transcription initiation at the malT promoter. Two lines of evidence supported this conclusion. First, mutation malTp1 rendered the expression of malT nearly independent of CAP, and CAP is known to act at the level of transcription (Pastan and Adhya, 1976; Simpson, 1980; Ullmann and Danchin, 1982). Secondly, malTpJ is located 12 nucleotides upstream from the transcription start, i.e., in the Pribnow box (Rosenberg and Court, 1979; Siebenlist et al., 1980) of the malT promoter. The replacement of a G by a T at this position results in the formation of a nucleotide hexamer (TACCTT) whose sequence is much closer to that of the "consensus" Pribnow box (TATAAT) than is that of the wild-type hexamer (GACCTT), the first T in the consensus sequence being one of the most conserved bases. We conclude that the malT promoter is intrinsicaly relatively weak, because of an inadequate Pribnowbox.The binding of RNA polymerase to this promoter would be enhanced by the CAP-cAMP complex. Mutation malTpJ would allow quasioptimum binding of RNA polymerase to the promoter,such that it could be only slightly enhanced by CAP and cAMP. Mutations similar to matTpl have been described in the lac (Arditi et al., 1973; Reznikoff and Abelson, 1978; Maquat and Remikoff, 1980) and ara (Hor- witzetal.,1980) systems.
The nature and location of the base change corresponding to matTp7 strongly suggest that this mutation lead stoan increase in the rate of translation initiation by allowing a more eficient binding of the ribosomes to malT mRNA. As a rule procaryotic cistrons are preceded by a short sequence, the "Shine and Dalgarno sequence", which is complementary to the 3' end of 16S rRNA and which is an important element of the ribosome binding site (Shineand Dalgarno,1975;Goldet al.,1981).Gene malT is preceded by such a sequence, which, however, is unusual in that it is mainly complementary to nucleotides 8-12 from the 3'end of 16S rRNA, while the sequence preceding other cistrons is usualy complementary to nucleotides located closer to the3' end of the rRNA (Figure 7).
Mutation matTp7 resulted in the creation of a GGAG sequence, which is complementary to the sequence extending from the fourth to the seventh nucleotide from the 3' end of 16S rRNA. The disance of this GGAG sequence from the initiation codon (nine nucleotides) is well within the range of distances found in other system (Gold ete al. 1981). In mutant malTp7, there is an almonst complete complementarity of the mRNA with a sequence extending from the fourth to twelfth nucleotide of the 3' end of 16S rRNA. Therefore, we conclude that the ribosome binding site located at the begginning of the malT cistron is normally rather weak, and that malTp7 resulted in the formation of a much stronger binding site, allowing more efficient initiation of translation. This conclusion is clearly compatible with the fact that malT expression is still controlled by CAP in a malTp7 mutant.

The finding that malT expression could be increased by mutations affecting either the promoter or the ribosome binding site was unexpected. Differential gene expression in bacteria is generally assumed to be mainly the result of different promoter eficiencies. Mutations leading to an increase in the expression of a particular gene are usually considered to be "up promoter" mutations. Mutations leading to an increased expression of the malT gene (Debarbouille and Schwartz, 1980) had been assumed to be located in the promoter. These mutations have not been mapped by DNA sequencing but, since they still allow malT expression to be controlled by CAP, they may very well be located, like malTp7, in the ribosome binding site.
A limitation of gene expression at the translation level offers one advantage for genes that are expressed at a very low rate: it provides a more regular synthesis of the gene product. In a wild-type strain there may be of the order of a hundred to a few hundred molecules of MalT polypeptide (Debar- bouilei and Schwartz,1979). If malT mRNA is translated at a high rate, the synthesis of one or two mRNA molecules per generation could suffice to produce such a concentration of MalT protein (Kenneland Riezman, 197).Statisticaly this would result in significant variations of the concentration of MalT protein in individual cels. Concomitant variations in the expression of the mal operons would occur (Nbarbouille and Schwartz, 1980) and this would presumably be undesirable. Such a situation can be avoided by having a higher rate of transcription, but a lower rate of translation. Similar reasoning could be applied to the case of gene CI in the X prophage. This gene is also transcribed at a reasonable rate, but poorly translated, presumably because the mRNA startswith AUG and therefore does not contain a Shine and Dalgarno sequence (Ptashneetal.,1976; Schereretal.,1980). In that case the importance of having as mooth synthesis of the gene product is even more obvious.
Mutations malTpl and malTp7 were recombined into a hybrid malT-lac operon, which contained a malT-lacZ hybrid gene as the promoter proximal gene, and lacY, as a distal gene. Mutation malTpl resulted in a parallel increase (- 5- fold) in the expression of the two genes, as would be expected for a mutation leading to an increase in transcription initiation.Mutation malTp7,on the other hand,led to a greater increase in the expression of the proximal gene (14-fold) than of the distal gene (3-fold). Therefore, the expression of a distal gene is not strictly correlated with the level at which the proximal gene is translated.A similar conclusion was recenfly drawn from studies of a lamB-lachybrid operon(Schwartzet al.,1981). Further investigations on this point may shed additional light on the mechanism of polarity in operon expression.
A strain that contained both matTpl and malTp7 cis with respect to another wise wild-type malT gene produced ~30 times more MalT protein than a wild-typestrain. For reasons that are not understood, this is somewhat lower than would be expected from a strictly cumulative efect of the two mutations. Unlike strains that harboured a malT gene on a multicopy plasmid (Raibaud and Schwartz, 1980) the double mutant did not show any obvious growth defect. It expressed the maltose operons constitutively and was,therefore, pheno-typicaly similar to previously described malT mutants (DNbarbouil& et al., 1978). In the later case, however, the mutations were located within the coding sequence and presumably affected the structure, rather than the synthesis, of MalT protein.

Materials and methods
Strains and media 
The strains of E. coli K12 used are listed in Table I. All growth media [complete medium (ML) and synthetic medium (M63)] were previously described (Chapon, 1982). Ampicillin was used at 100 pg/ml in solid or liquid media. Tetracycline, unless otherwise stated, was used at 5 plg/ml in liquid media and 10 pg/ml in solid media.  

Enzyme asays
B-Galactosidase (B-D-galactoside galactohydrolase, EC 3.2.1.23) was assayed according to Miller (1972). The lacY produc t(3-galactosidepermease) was assayed by measuring the accumulation of C4C-labeled thiomethyl,B-D-galactoside (Rickenberg et al., 1956). The plate test for detection of clones synthesizing amylomaltase constitutively was as described by Debarbouille et al.(1978).

Transfer of the mutations from the chromosome to the plasmid
The techniques used were almost exactly as described by Raibaud and Schwartz (1980). Genetic mapping had shown that the malTp mutations fail to recombine with deletion malAA510 (Chapon, 1982). They were transferred from the chromosome to plasmid pOMI (RaibaudandSchwartz,1980) using the selection described in Figure1. However,the scheme shown in the figure represents a simplification of the actual events, especialy because the cels contain several copies of the plasmids. Most of the plasmids extracted from the TetR clones stil carried the deletion. To purify those that harboured the malTp mutation, the plasmids extracted from the TetR clones were used to transform a malT derivative of C600. The TetR transformants were selected on ML agar containing 20 ug/ml tetracycline. (With lower concentrations of tetracycline the transformants stil contained a mixture of plasmids carrying the malTp mutation and plasmids carrying the deletion. This is probably the result of the toxicity of plasmids that express both malT and the malP-tet hybrid operon).The plasmids extracted from the TetR transformants gave the same digestion pattern with HincII as observed with pOMI. The presenc eof the malTp mutations on the plasmids was verified by transfering the mutations back to the chromosome, by a procedure the reverse of that shown in Figure1 with selection for a Mal+ phenotype. Since the curing of the resulting strains was usualy dificult, the phenotype,i.e., partialy constitutive a mylo-maltasesynthesis (D&barbouile et al.,1978) was studied after transducing the malT gene into a plasmid-free strain(pop2150,TableI). 

The 5'-endlabeledi DNA fragments were prepared esentialy as described by Debarbouile et al.(1982).The DNA was sequenced by the method of Maxam and Gilbert (1980). The end-labelled fragments, obtained after the chemical degradations,were analysed by electrophoresis on thin gels(0.3mm) containing 7% acrylamide (SangerandCoulson,1978).
Acknowledgements
I am particularly indebted to Michel Debarbouille who introduced me to the techniques of DNA sequencing and reverse transcriptase mapping. I thank Olivier Raibaud for many useful suggestions and Maxime Schwartz for his constant interest in this work and his help in preparing the manuscript. This work was supported by grants from the CNRS (LA 270) and from the DGRST (no. 80 7 01 82)