The MalT-dependent and malZ-encoded maltodextrin glucosidase of Escherichia coli can be converted into a dextrinyltransferase by a single mutation.

malZ is a member of the mal regulon. It is controlled by MatT, the transcriptional activator of the maltose system. MalZ has been purified and identified as an enzyme hydrolyzing maltotriose and longer maltodextrins to glucose and maltose. MalZ is dispensable for growth on maltose or maltodextrins. Mutants lacking amylomaltase (encoded by malQ), the major maltose utilizing enzyme, cannot grow on maltose, maltotriose, or maltotetraose, despite the fact that they contain an effective transport system and MalZ. From such a malQ mutant a pseudorevertant was isolated that was able to grow on maltose. The suppressor mutation was mapped in malZ. The mutant gene was cloned. It contained a Trp to Cys exchange at position 292 of the deduced protein sequence. Surprisingly, the purified mutant enzyme was still unable to hydrolyze maltose as was the wild type enzyme, while both were able to release glucose from maltodextrins. However, the mutant enzyme had gained the ability to transfer dextrinyl moieties to glucose, maltose, and other maltodextrins. Thus, it had gained an activity associated with amylomaltase. It was the MalZ292-associated transferase reaction that allowed the utilization of maltose. In addition, we discovered that mutant and wild type enzyme alike were highly active as gamma-cyclodextrinases.

The Escherichia coli maltose system (1,2) contains two enzymes that are necessary for the utilization of maltose and maltodextrins. Maltotriose, after having been taken up by the high affinity and binding protein-dependent ABC (ATP binding cassette) transport system (3) is recognized by amylomaltase (encoded by malQ) (4), the reducing end glucose is released and the maltosyl residue is transferred to another maltotriose molecule thus forming maltopentaose (5,6). The repetition of this cycle leads to the formation of long maltodextrins and free glucose which, after phosphorylation by glucokinase enters glycolysis. Maltopentaose and longer maltodextrins are recognized by maltodextrin phosphorylase (encoded by malP) (7,8) which, by phosphorolysis, releases the nonreducing end glucose as glucose 1-phosphate. Thus, the final products of maltodextrin metabolism by these two enzymes are glucose and glucose 1-phosphate.
The degradation of maltose, the smallest member of maltodextrins, also requires amylomaltase, and malQ mutants are Mal Ϫ . However, amylomaltase does not recognize maltose as glucosyl donor, only as maltodextrinyl acceptor (6). Therefore, in order to metabolize maltose, amylomaltase requires a maltodextrin primer with the minimal size of maltotriose. Within the cell, the required maltodextrin primer can originate from the degradation of glycogen or from the action of an as yet uncharacterized maltose/maltotriose phosphorylase with glucose and glucose 1-phosphate as starting material (9).
There are two more enzymes, members of the mal regulon, that are not essential for the metabolism of maltose and small maltodextrins. One is a periplasmic ␣-amylase encoded by malS (10,11). This enzyme hydrolyzes larger dextrins in the periplasm preferentially to maltohexaose (12) which can then be transported by the maltose/maltodextrin transport system. The other is a cytoplasmic enzyme encoded by malZ (13,14) whose function in maltodextrin metabolism is unclear. The enzyme degrades linear maltodextrins up to a chain length of 7 glucose units but not maltose. It sequentially cleaves glucose from the reducing end of the maltodextrin chain. Even though it can hydrolyze maltotriose to glucose and maltose, malQ mutants cannot grow on maltotriose or maltotetraose even when malZ is overexpressed (14).
The endogenous formation of maltotriose required for maltose utilization is also an important aspect of the regulation of the mal system. All mal genes are under the control of MalT, the positive activator of the system (15). In vitro MalT-dependent mal gene expression requires the presence of maltotriose (16). Cells growing in the presence of maltose or any other maltodextrin will result in the elevated expression of the mal system even though only maltotriose is the effective internal inducer. We have argued that the increased formation of endogenous maltotriose during exogenous induction by maltodextrins is driven by glucose and glucose 1-phosphate, the degradation products of maltodextrin metabolism. This reaction is most likely catalyzed by maltose/maltotriose phosphorylase, the same enzyme that is required for the maltodextrin primer synthesis mentioned above (9).
In an attempt to identify the postulated maltose/maltotriose phosphorylase, we reasoned that the action of this enzyme should be reversible, and it should therefore be able to split maltose to glucose and glucose 1-phosphate. Thus, malQ mutants which are Mal Ϫ should be able to grow again on maltose after a mutational event resulting in the overproduction of the postulated maltose/maltotriose phosphorylase. Among the phenotypic revertants in such a selection, we found one mutation that did not map in malQ. In this paper, we report the analysis of this second site revertant. We found that the mutation had occurred in malZ. We cloned and sequenced the mutant malZ (malZ292). We purified the encoded enzyme and characterized its enzymatic activity. We present the explanation for the ability of the mutant enzyme to support growth on maltose by its acquisition to transfer dextrinyl residues to maltose, followed by the consecutive hydrolysis of glucose from the transfer product.

EXPERIMENTAL PROCEDURES
Construction of Strains-Derivatives of E. coli K12 strains were used throughout this study. They were constructed by P1 vir transduction (17) and are listed in Table I. To isolate the malZ292 mutation, strain HS3166, a maltose constitutive malQ am mutant, was grown in minimal medium A (MMA) 1 (17) containing 0.4% glycerol. 2 ϫ 10 10 cells were plated on MMA with 0.2% maltose. Colonies were tested for the reversion of the malQ am mutation by transducing them with a P1 lysate prepared on POP4064 (HfrG6 malQ::Tn10). Transductants were selected for tetracycline resistance (Tet R ) and screened for a Mal ϩ phenotype.
DNA Methods-Chromosomal DNA was isolated as described by Silhavy et al. (18). Plasmid DNA was prepared by the alkaline-SDS method according to Birnboim and Doly (19). Other standard recombinant DNA techniques are from Sambrook et al. (20).
Construction of pRP100 and Derivatives-Chromosomal DNA of SFC1 was digested with BamHI and PstI. The 10 -14-kb fraction was eluted from an agarose gel and ligated into pACYC177 that had been digested with the same enzymes. Strain RP90 was transformed with this ligation mixture. Colonies were selected for kanamycin resistance and screened for the ability to derepress alkaline phosphatase on G ϩ L medium (21) with 0.1 mM P i and 0.2% glucose as carbon source and 40 g/ml 5-bromo-4-chloro-3-indolyl phosphate (X-P) as indicator for alkaline phosphatase. All the clones which were able to derepress alkaline phosphatase were also able to grow on MMA-plates containing 0.2% maltose. One of these plasmids, pRP100, contained a chromosomal insert of about 12 kb and was used for further studies. pRP102 was obtained by deleting two MluI fragments (together 6 kb) from pRP100. For the construction of pRP112, pRP102 was digested with SacII, and the resulting fragments were blunt-ended with T4 polymerase. In a second step these fragments were digested with MluI. A 2.1-kb MluI/ SacII fragment was isolated and ligated in pDHB32 that had been digested with MluI and PshAI. The resulting plasmid pRP112 contained a 2.1-kb chromosomal fragment encoding the entire mutant malZ292 gene (Fig. 1). For the construction of pRP117, pRP112 was digested with HindIII and SnaBI, and a 742-bp fragment containing the malZ292 mutation was eluted from an agarose-gel. Plasmid pST4 (14) was also digested with HindIII and SnaBI. The above 742-bp fragment was ligated in the opened pST4 plasmid. After transformation of RP90 with the ligation mixture, all clones were able to grow on MMA plates containing 0.2% maltose. The construction of pRP118 and pRP119, the expression plasmids for malZ and malZ292, was done by polymerase chain reaction technique using two primers (ACAGGGGACATATGAT-GTTAAATGC) and (TTCGCTGGTACAGCGCG). A 1421-bp fragment of malZ was amplified in this way using pRP112 as template. The amplified fragment was digested with NdeI, and the resulting 1020-bp fragment was ligated in the heat-inducible expression vector pCYTEP1 (22). The correct orientation of the insert was confirmed by restriction analysis. The 672-bp BglII/EcoRI fragment of this vector was exchanged against a 1491-bp BglII/MunI fragment from pRP112 or pRP114 completing malZ292 and malZ, respectively. pRP112 and pRP114 are equivalent with the exception of the malZ292 mutation.
DNA Sequencing-To sequence the malZ and the malZ292 genes, the plasmids pST4 and pRP112 were used. Plasmid preparation was done with the Quiagen Kit (Diagen GmbH Hilden, Germany). Doublestranded DNA was denatured (23) and sequenced by the chain termination method (24) with the Sequenase kit (U. S. Biochemical Co.) using seven primers obtained from the MWG-BIOTECH Co.
Purification of MalZ and MalZ292-Strain RP96 harboring either pRP118 or pRP119 was grown at 28°C in 2 liters of LB (17) to an optical density at 578 nm (A 578 ) of 1. The culture was shifted to 42°C for 2 h, harvested by centrifugation, washed in 50 mM Hepes, pH 7.0, and resuspended in 35 ml of the same buffer. The cells were disrupted by one passage through a French pressure cell at 16,000 p.s.i. DNase was added followed by incubation for 30 min. Inclusion bodies were collected by centrifugation (12,000 ϫ g for 10 min). The pellet was solubilized in 10 ml of 50 mM Hepes containing 4 M guanidinium HCl and 200 mM dithiothreitol. The protein concentration was adjusted to 1 mg/ml (total volume 100 ml) and put into a dialysis bag with 300 ml external volume of the same buffer containing 4 M guanidinium HCl. Renaturation was achieved by slowly (during 12 h) diluting the outside dialysate with the 5-fold volume of MMA at 4°C. Subsequently, the protein solution was dialyzed twice against 2 liters of MMA overnight. Remaining protein precipitate was removed by centrifugation (18,000 ϫ g for 30 min). The solution was precipitated with 40% saturated (NH 4 ) 2 SO 4 and cleared by centrifugation. The supernatant was precipitated with 60% (NH 4 ) 2 SO 4 . The precipitate was collected by centrifugation, dissolved in 50 mM sodium phosphate buffer, pH 7.0, and dialyzed against the same buffer. The protein was further purified in 2-ml portions by fast protein liquid chromatography anion exchange column chromatography (Mono Q, Pharmacia Biotech Inc.) in 50 mM sodium phosphate buffer, pH 7.0. The column was eluted with 20 ml of a linear NaCl gradient (0 -300 mM) in the same buffer.
Transport of [ 14 C]Maltose-Cultures were grown logarithmically in MMA containing 0.4% glycerol as carbon source to an A 578 of 0.5. When indicated, 1 mM maltose was added 2 h previously. The cultures were harvested, washed three times with MMA, and resuspended in the same medium to an of A 578 of 0.5 (glycerol-grown cultures) or 0.05 , and washed with 10 ml of MMA. The radioactivity was determined in a liquid scintillation counter (Beckman LS 1801). The rate of uptake was determined from the linear increase in the accumulated radioactivity and is given as picomoles of substrate taken up per 10 9 cells per min.
Preparation of Labeled Maltodextrins-The reaction mixture (250 l of 50 mM Tris-HCl, pH 7.0) contained 100 mM maltotriose, 0.4 mM [ 14 C]glucose (5 Ci), and 0.5 mg of protein of crude extract obtained from strain ME469 harboring plasmid pCHAP113 (4). The strain had been grown in LB containing 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside. The mixture was incubated at room temperature for 1 h and subsequently heated in boiling water for 5 min. Under these conditions, the reaction has reached equilibrium. The suspension was clarified by centrifugation, and the supernatant was applied on Whatman 3MM paper for descending chromatography with butanol:ethanol:water (5: 3:2, v/v/v) as solvent. After drying, the paper was autoradiographed for 24 h (Fig. 7). Paper stripes containing the radioactive compounds were eluted with water and once rechromatographed with the same system. The products were again identified by autoradiography, eluted with water, and lyophilized. They were used in enzymatic digests without further purification.
Analytical Techniques-For testing the MalZ-dependent hydrolysis products, purified MalZ or MalZ292 (approximately 60 g/ml, final concentration) was incubated with maltodextrins (in 50 mM sodium phosphate buffer, pH 7.0) at room temperature for the time periods indicated. The standard assay was carried out in 100 l, containing maltodextrins at a final concentration of 10 mM (maltose) to 3.
After drying, the TLC plates were autoradiographed, with times of exposure ranging from 4 days to 2 weeks. The unlabeled oligosaccharide spots were visualized by dipping the plate in methanol containing 2% concentrated H 2 SO 4 followed by drying and charring for 10 min at 170°C.
MalZ activity was assayed by following the formation of p-nitrophenol from p-nitrophenyl maltoside (PNG2) as described previously (14). Units of enzymatic activity are given in micromoles of PNG2 hydrolyzed per min at room temperature.

RESULTS
Isolation of a Pseudorevertant from a malQ am Mutant-malQ mutants are defective in amylomaltase and cannot grow on maltose. In addition, when grown on a different carbon source, they are sensitive to maltose. We plated the malQ am mutant HS3166 on minimal maltose plates and selected for growth. Phenotypic revertants could easily be obtained. To screen for those that were not true revertants in malQ, we transduced them to Tet R with a P1 lysate obtained from a malQ::Tn10 strain. Most of the Mal ϩ revertants became again Mal Ϫ indicating that these Mal ϩ revertants were indeed true revertants. One strain, named SFC1, remained Mal ϩ after the introduction of malQ::Tn10 and was considered further. Using the collection of P1 lysates of mapped Tn10 insertions (25), we found that the second site reversion had occurred near the region of minute 9 on the chromosome. A P1 transduction using the phoB::Tn5 mutant BW10320 as donor and tetracycline resistance as selection revealed that the Mal ϩ phenotype was linked to phoB (78% cotransduction frequency). This raised the possibility that the Mal ϩ phenotype was caused by a mutation in malZ, a mal gene located next to phoB (26).
Cloning and Sequencing the Mutated malZ Gene-The high cotransduction frequency between the second site mal ϩ mutation in SFC1 and phoB::Tn5 prompted us to use the latter as selection to clone the gene in which the mutation had occurred. Chromosomal DNA of SFC1 was digested and ligated into plasmid pACYC177. The phoB::Tn5 derivative of strain HS3166 (malQ am ) was transformed with the ligation mixture. Colonies were selected for resistance against kanamycin, and the ability to derepress alkaline phosphatase was screened. All these clones were able to grow on maltose. pRP100, the plasmid in one of these clones, contained a chromosomal insert of about 12 kb. With the assumption that the gene in question was malZ, we cloned a 2.1-kb MluI/SacII fragment containing malZ in pDHB32 yielding pRP112 (Fig. 1). This plasmid still carried the mal ϩ second site mutation but was no longer able to complement a phoB mutation. The insert was sequenced in its entire length and verified the identity of its open reading frame as MalZ. The mutant gene contained a G to T exchange in comparison to the wild type malZ sequence resulting in the exchange of Trp to Cys at position 292 of the MalZ polypeptide chain. 2 Purification of Mutant and Wild Type MalZ Proteins-In order to study the enzymatic properties of the mutant MalZ enzyme free of activities that might be caused by other maltodextrin-hydrolyzing activities, the purification of the mutant and wild type MalZ was necessary. For this purpose, malZ and malZ292 were cloned into the plasmid pCYTEXP1 (22) allowing the expression of both genes under temperature-sensitive repressor control. In both cases, induction resulted in the formation of inclusion bodies. This material was isolated by differential centrifugation followed by solubilization in 4 M guanidinium HCl. Renaturation was achieved by slow dilution in phosphate buffer. Gel electrophoretically homogeneous protein was obtained by anion exchange chromatography in a Mono Q column. The analysis of the different purification steps by SDS-polyacrylamide gel electrophoresis is shown in Fig. 2. All tests were done with protein purified in this manner. To ensure that the observed properties were not due to the denaturationrenaturation treatment, the key experiments (transfer reaction, cyclodextrinase activity) were repeated with crude extract containing soluble native MalZ as well as MalZ292. No differences to the preparation obtained from inclusion bodies were observed.
MalZ292 Is Not Only a Hydrolase as the Wild Type MalZ but Also a Transferase-From the isolation of the malZ292 mutation as a phenotypic Mal ϩ revertant of a malQ am mutant, it seemed likely that the MalZ mutant enzyme would have acquired the ability to hydrolyze maltose, a property that is absent in the wild type MalZ enzyme. However, maltose was not hydrolyzed by MalZ292 even after long incubation times (data not shown). Maltopentaose was hydrolyzed (mainly to glucose and maltose) by the wild type as well as the mutant enzyme (Fig. 3A). However, the presence of trace amounts of 14 C-labeled glucose or 14 C-labeled maltose in the hydrolysis mixture revealed that the mutant MalZ enzyme was able to transfer dextrinyl residues onto the labeled glucose and maltose while the wild type enzyme was not, or only to a minor extent (Fig. 3B). Most significantly, 14 C label originally contained in maltose was released as [ 14 C]glucose during the experiment. In the attempt to detect the initial transfer product arising with the mutant enzyme, we tested maltose, maltotriose, maltotetraose, and maltopentaose as glucosyl donors with 14 C-labeled glucose or 14 C-labeled maltose as acceptors (Fig. 4). As expected, maltose was not used as a glucosyl donor. When [ 14 C]glucose was the acceptor and unlabeled maltotriose the donor, the initial labeled product was [ 14  donor. This first product will thus contain two consecutive 14 C-labeled glucose residues at the reducing end. This product itself is a good substrate of MalZ and will lead to the quick release of [ 14 C]glucose as well as a 14 C-labeled dextrin that is smaller by one glucose moiety. With unlabeled pentaose as donor, the results are less clear since the products formed were not sufficiently separated by the TLC technique. It is noteworthy to emphasize that the MalZ292 enzyme, aside from its increased activity as a dextrinyl transferase, still shows the same activity as the wild type enzyme in its function as a maltodextrin glucosidase, i.e. in the net formation of glucose from maltodextrins.
MalZ Is a ␥-Cyclodextrinase-In our previous publication on the properties of MalZ, we reported that the enzyme only hydrolyzed linear maltodextrins up to seven glucose residues in chain length. The two cyclodextrins tested, ␣and ␤-cyclodextrin with a length of six and seven glucosyl residues, were not hydrolyzed (14). In testing the substrate specificity of the mutant MalZ enzyme, we found that it readily hydrolyzed ␥-cyclodextrin (containing eight glucosyl residues) and to a very minor extent ␤-cyclodextrin but not ␣-cyclodextrin. The same was true for the wild type MalZ enzyme (Fig. 5). Following the time-dependent appearance of the products, we first observed the formation of linear maltooctaose, followed by linear maltoheptaose. The same products and not the starting material (␥-cyclodextrin) became labeled from the trace amounts of [ 14 C]glucose present as acceptor. Only after some time, glucose, maltose, and small amounts of maltotriose became labeled (Fig.  6). Apparently linear maltooctaose and maltoheptaose are rather poor substrates. Only after the removal of glucose or maltose moieties from the reducing end of these dextrins does the subsequent hydrolysis proceed rapidly. Again, the transfer reaction occurred in significant amounts only when the mutant MalZ292 enzyme was used but not with the wild type MalZ FIG. 1. Cloning of malZ292. a, region of the E. coli chromosome around minute 9 of the genetic map that is cotransducible with the malQ am suppressor mutation. b, plasmid pRP100 that had been selected for its ability to complement a phoB::Tn5 mutation and which carried the malZ292 mutation. c and d, subcloning of malZ292 for minimal size. The letters above the lines show the restriction sites: B, BamHI; M, MluI; P, PstI, Ps, PshAI; S, SacII. ␥-Cyclodextrinase of E. coli enzyme (data not shown). We conclude that the hydrolysis of ␥-cyclodextrin by the mutant as well as the wild type enzyme occurs by opening the ring followed by the consecutive release of glucose or maltose from the reducing end of the linear dextrin finally yielding glucose and maltose.
Determination of Glucose after MalZ-dependent Hydrolysis of Maltodextrins-The qualitative appearance on TLC plates of about equimolar amounts of glucose and maltose after the hydrolysis of ␥-cyclodextrin, obviously incompatible with a consecutive release of glucose residues after opening the circular dextrin, prompted us to reanalyze the amount of glucose released from different dextrins after end point hydrolysis with the MalZ enzyme. These data are shown in Table II. A consecutive hydrolysis of glucosyl residues from the reducing end of the maltodextrins would result in the liberation of 1, 2, 3, and so forth glucose from maltotriose, maltotetraose, maltopentaose, and so on. Whereas maltotriose and small dextrins followed this pattern, from increasingly longer dextrins too little glucose was released. With heptaose as substrate instead of 5 eq of glucose, only three were released. From ␥-cyclodextrin instead of six glucose molecules, only four were released. Thus, it was likely that MalZ not only liberates glucose but also maltose or maltotriose from longer maltodextrins.
When the mutant MalZ enzyme was used in this assay, the amount of glucose released was significantly higher with all maltodextrins tested. We interpret this as the consequence of the maltodextrinyl transfer reaction associated with the mutant enzyme. This would result in the transfer of dextrinyl moieties onto maltose and subsequent release of glucose. This reaction is essentially the reason why the malZ292 mutation in a malQ background gives rise to maltose utilization.
Hydrolysis of Maltotriose and Maltotetraose 14 C-labeled at the Reducing End Glucose Residue-In our previous publication on the mechanism of the MalZ enzyme, we had concluded that the enzyme is a glucosidase releasing glucose consecutively

␥-Cyclodextrinase of E. coli
from the reducing end of maltodextrins (14). We showed that maltotriose 14 C-labeled exclusively at the reducing end glucose molecule yielded only 14 C-labeled glucose but not 14 C-labeled maltose upon treatment with the MalZ enzyme. We repeated these experiments now not only with maltotriose but also with maltotetraose 14 C-labeled only in the reducing end molecule.
These dextrins were synthesized by the transferase reaction of amylomaltase. Unlabeled maltotriose (100 mM) was incubated with [ 14 C]glucose (0.4 mM) and amylomaltase. The products formed were separated by paper chromatography, and the results are shown in Fig. 7. According to the established mechanism of amylomaltase action (6), these sugars are labeled exclusively in the reducing end glucose residue. After rechromatography, maltotriose and maltotetraose were used in the MalZ-dependent hydrolysis reaction. The results are shown in Fig. 8. We now conclude that from maltodextrins longer than maltotriose not only glucose but also maltose can be released to a small extent.

MalZ292-dependent in Vivo Maltose Utilization Occurs in the Absence of Amylomaltase and Maltodextrin Phosphorylase-
The properties of the MalZ292 enzyme suggested that the utilization of maltose as carbon source would, aside from the requirement of a dextrin primer, not necessitate the presence of an additional enzyme. Thus, in a net reaction, glucose would be formed from maltose by MalZ292. It was therefore necessary to demonstrate that MalZ292 would allow growth on maltose in the absence of amylomaltase (the malQ-encoded enzyme) and maltodextrin phosphorylase (the malP-encoded enzyme).
Surprisingly, a malT ϩ malP::Tn10 mutant (strain RP97) lacking maltodextrin phosphorylase and (by the polar effect of  a The actual initial substrate concentration was between 10 mM (maltotriose) and 3 mM (␥-cyclodextrin). The results were standardized for an initial substrate concentration of 10 mM.

␥-Cyclodextrinase of E. coli
the Tn10 insertion on malQ) also amylomaltase and carrying malZ292 as a chromosomal mutation utilized maltose much better than strain RP98 (malT ϩ malQ::Tn10 malP ϩ ) with malZ292 as a chromosomal mutation. Careful analysis with the purest available maltose (Merck) revealed that strain RP98 only grew overnight to an A 578 of about 0.3-0.5. After a long incubation time (between an additional 10 and 20 h), growth resumed but was due to malQ ϩ reversions. In contrast, strain RP97 (malP::Tn10 malQ malZ292) grew well and was fully outgrown overnight on the same purest maltose preparation (Merck). In addition, the sterile filtered spent medium of strain RP98 (malQ::Tn10 malP ϩ malZ292), harvested at a time when growth had stopped at A 578 of 0.4, allowed full growth of strain RP97 (malP::Tn10 malQ malZ292). As tested by TLC, the spent medium still contained large amounts of maltose. This indicates that endogenously produced maltodextrin needed for the MalZ292-mediated utilization of maltose will be partially eliminated by maltodextrin phosphorylase. Apparently less pure maltose preparations contain small quantities of maltodextrins allowing the MalZ292-mediated utilization of maltose even in the presence of maltodextrin phosphorylase.
Lack of Phosphoglucomutase (Encoded by pgm) Does Not Prevent MalZ292-dependent Utilization of Maltose-We found that the introduction of a pgm mutation (lacking phosphoglucomutase) into a malQ malZ292 strain (RP91) did not prevent the utilization of maltose; under these conditions, malP ϩ encoding maltodextrin phosphorylase no longer prevented growth in the pgm mutant even on pure maltose. This again demonstrated firstly that in contrast to the amylomaltase/maltodextrin phosphorylase-mediated pathway the "usable glucose unit" produced from maltose in the malZ292 mutant was not glucose 1-phosphate but glucose. Secondly, it showed that the inability to utilize glucose 1-phosphate via phosphoglucomutase counteracted the maltodextrin-degrading activity of maltodextrin phosphorylase. Possibly, maltodextrin phosphorylase under conditions of excess glucose 1-phosphate even catalyzes the reverse reaction, the synthesis of maltodextrins.
The malZ292 mutation was isolated as a second site Mal ϩ revertant from a malQ mutant but obviously not on pure maltose. We determined the growth rate of the malQ malZ292 mutant on maltose in comparison to a malQ malZ ϩ strain. While the malZ ϩ strain did not grow at all, we did notice that the malZ292 derivative grew on maltose at different growth rates depending on the brand of maltose that was used as carbon source (Table III). The introduction of a pgm mutation lacking phosphoglucomutase activity, abolished the difference in the growth rates between the pure and less pure maltose and allowed reasonable growth rates and final cell densities in both cases (Table III). It was somewhat surprising that the rate of growth on maltose in a malQ malP pgm malZ292 strain (RP93) was dependent on the copy number of the malZ292 gene, the rate of growth becoming considerably higher when malZ292 was plasmid-encoded than when chromosomally encoded.

The MalZ292-mediated Utilization of Galactose in a pgm Mutant Requires the Presence of malP-encoded Maltodextrin
Phosphorylase-Recently we had concluded that growth on galactose in a pgm mutant (27) required the presence of all maltose utilizing enzymes, MalP, MalQ, and MalZ. Mutations in any of the genes encoding these enzymes abolished the ability to grow on galactose and to develop the galactose blue phenotype (9). It was therefore of interest whether or not the malZ292 mutation in a malQ pgm mutant would allow growth on galactose. Strain RP91 (malZ292 malQ pgm) could still grow on galactose, albeit slowly. However, RP99 (malQ malP pgm) harboring plasmid-encoded malZ292 (pRP117) did not grow at all on galactose. Thus, the formation of maltodextrins from the galactose-derived glucose 1-phosphate (via the postulated maltose/maltotriose phosphorylase involved in inducer synthesis) and their subsequent degradation to glucose by the MalZ292 enzyme is too slow to support growth on galactose. We therefore conclude that galactose utilization in a malZ292 pgm mutant appears to require maltodextrin phosphorylase, the malP gene product, whereas maltose utilization via MalZ292 does not.
The Glycogen-derived Production of Maltodextrins Is Not Essential for the MalZ292-dependent Utilization of Maltose-The introduction of a glgA mutation that prevents glycogen formation did not alter the ability of the malQ pgm malZ292 mutant to grow on maltose (strain RP93). This demonstrates once again that aside from the glycogen-dependent pathway a second origin of maltodextrins needed as primers for the MalZ292-dependent maltose utilization must exist. Since malQ pgm or malQ pgm glgA strains are no longer constitutive for the expression of the maltose system (see Table IV), the endogenous synthesis of maltotriose (in the absence of external maltose) must be slow. It is certainly slower than the production of maltotriose from glycogen.
The expression of the maltose system is controlled by the level of maltotriose, the endogenous inducer. Obviously, mal-  ␥-Cyclodextrinase of E. coli totriose can be produced and degraded by several enzymes. Thus, it is informative to analyze the state of mal gene expression in several mutants in the presence and absence of exogenous maltose. Expression was measured by maltose transport activity (Table IV). The constitutivity of a malQ mutant is somewhat less in the presence of the malZ292 mutation while the presence of maltose in such a strain increases the expression more than in a malZ ϩ strain (SFC1 versus HS3166). As expected, the introduction of a glgC or a pgm mutation in SFC1 abolishes the constitutive expression, but does not prevent induction by exogenous maltose. MalZ292 Reduces the Growth-inhibitory Effect of Maltose in a malQ Mutant-Another phenotype associated with the malZ292 mutation is its protective function against the growth inhibition by maltose in a malQ mutant. It has been known for a long time that growth of malQ mutants is sensitive to maltose (28). This can easily be observed on McConkey or EMB indicator plates containing maltose. Under these conditions, Mal Ϫ mutants arise frequently that shut off the influx of maltose by the acquisition of either malK or malT mutations (1). This maltose-sensitive phenotype largely disappears when the malQ mutant also carries the malZ292 mutation.
The Lack of the MalZ Enzyme Exhibits a Maltose Phenotype Only in pgm Mutants Lacking Phosphoglucomutase-Up to now, no maltose phenotype has been associated with the lack of the MalZ enzyme in an otherwise wild type strain. The function of this enzyme only becomes apparent in a pgm mutant. There, of the two final products of maltodextrin metabolism, glucose and glucose-1-phosphate, only glucose can be used as carbon source. The pgm mutant (strain RP101) can still grow on maltose. However, the combination of a malZ and a pgm mutation results in strongly reduced growth on maltose. Obviously, the function of MalZ is to increase the proportion of glucose in the final products of maltose utilization.
The finding that MalZ292 as well as the wild type MalZ enzyme is an effective ␥-cyclodextrinase raised the question whether or not this activity is of physiological significance. Strain MC4100, our control strain as a standard laboratory Mal ϩ strain, cannot grow on ␥-cyclodextrin.

DISCUSSION
In this publication we report the mutational alteration (Trp to Cys at position 292 of the polypeptide chain) of MalZ that results in the effective utilization of maltose when present in a mutant that lacks amylomaltase, the key enzyme of maltose utilization in E. coli (1). MalZ had previously been identified as an enzyme cleaving glucose from the reducing end of maltodextrins with maltotriose as the smallest substrate. The wild type MalZ enzyme is apparently not involved in maltose utilization since mutants lacking malZ still grow normally on maltose and strains harboring the wild type malZ gene but defective in malQ cannot grow on maltose (29). However, in pgm mutants which still can grow on maltose (27), the function of the malZencoded enzyme becomes apparent. Growth on maltose is strongly reduced. Thus, MalZ increases the ratio of glucose over glucose 1-phosphate, the immediate products of maltose utilization.
The mutant enzyme MalZ292 still exhibits the same activity as the wild type MalZ enzyme. Surprisingly, maltose itself was not hydrolyzed by the mutant enzyme. The only difference that we could detect was the ability of the mutant enzyme to transfer dextrinyl residues originating from maltotriose and larger dextrins onto maltose. This indicates that it is the transfer reaction onto maltose that allows maltose utilization. The transfer reaction observed with the mutant MalZ enzyme is reminiscent of the action of amylomaltase, the gene product of malQ. This enzyme will disproportionate any given maltodextrin longer than maltose into glucose and a series of maltodextrins in such a way that the number of glycosidic linkages will remain constant. Thus, amylomaltase is exclusively a maltodextrin transferase but not a hydrolase. For every glucose released (the apparent hydrolysis reaction), it will form a new glycosidic linkage by producing a longer dextrin.
Even though the hydrolysis and the transfer reaction observed with the mutant MalZ enzyme is formally the same as in amylomaltase, there is still a basic difference between the two enzymes. In the mutant MalZ, hydrolysis of glucose is not obligatorily coupled to the transfer reaction, and net hydrolysis of maltodextrins to glucose and maltose is still the predominant reaction. Nevertheless, with a continuous supply of small amounts of dextrins (maltotriose and larger), maltose can be degraded to glucose by the mutant MalZ enzyme. This does not require the presence of an additional enzyme. In contrast, when maltose is degraded to glucose by the "normal" amylomaltase-mediated pathway, aside from the requirement of a maltodextrin primer, the action of maltodextrin phosphorylase, the malP product, is needed to remove the accumulation of longer dextrins (produced by amylomaltase) by producing glucose 1-phosphate that enters glycolysis after phosphoglucomutasemediated transformation into glucose 6-phosphate.
The properties of the mutant MalZ enzyme again necessitate postulating an endogenous production of maltodextrins that can act as a dextrinyl donor in the MalZ292-mediated utilization of maltose. In the malQ mutant, the major source of these dextrins including maltotriose, the inducer of the system, is clearly glycogen (9,30). However, also the pgm or the pgm glgA derivative of the malQ malZ292 strain can still grow on maltose. These strains do not contain detectable glycogen and are not constitutive for the maltose system even though they still can be induced by maltose. Thus, it is obvious that there exists a second pathway for the synthesis of endogenous maltodextrins that do not originate from glycogen. We have previously postulated the existence of a maltose/maltotriose phosphorylase that would produce maltose from glucose and glucose 1-phosphate and additionally maltotriose from maltose and glucose 1-phosphate (9). Since phosphorylases are reversible, maltose plus phosphate would produce glucose 1-phosphate which then could give rise to the synthesis of small maltodextrins needed for the MalZ292-mediated maltose utilization in the malQ mutant background. The observation that pgm mutants in this background exhibit an increased rate of maltose ␥-Cyclodextrinase of E. coli utilization is consistent with this picture. Glucose 1-phosphate produced from maltose (by the postulated maltose/maltotriose phosphorylase) would not be removed by phosphoglucomutase (forming glucose 6-phosphate followed by glycolysis). A testable prediction would therefore be that malQ pgm mutants when exposed to maltose will have an elevated level on glucose 1-phosphate even in the absence of maltodextrin phosphorylase. Clearly, the postulated maltose/maltotriose phosphorylase cannot be very active. Otherwise, this enzyme alone should give rise to growth on maltose in the absence of any other mal enzymes.
In the course of this investigation, we observed that the MalZ enzyme, wild type as well as MalZ292 mutant, was able to hydrolyze ␥-cyclodextrin, but not ␣-cyclodextrin, and ␤-cyclodextrin only to a minor extent. Several cyclo-maltodextrinases have been isolated in the past and identified as a special type of maltodextrin hydrolase. They exhibit molecular weights of 66,000 -72,000 and differ from ␣-amylases, to which they exhibit sequence homology (including their conserved motifs), by their weak activity in hydrolyzing starch. Generally, they exhibit transglycosylating activity and some of them are pullulanases hydrolyzing an ␣-(136) linkage (31)(32)(33). MalZ, even though exhibiting ␥-cyclodextrinase activity, carries latent transglycosylating activity that becomes prominent only after mutation. We could not detect any hydrolyzing activity on pullulan consistent with the claim that E. coli does not contain pullulanase (34). At present, the physiological role, if any, of the ␥-cyclodextrinase activity of MalZ is unclear. E. coli is unable to transport or to grow on ␥-cyclodextrin.