Site-specific Mutagenesis of Residues in the Escherichia coli Mannitol Permease That Have Been Suggested to Be Important for Its Phosphorylation and Chemoreception Functions*

The Escherichia coli mannitol permease is an inte- gral membrane protein that catalyzes the concomitant transport and phosphorylation of D-mannitol and also acts as the chemoreceptor for chemotaxis of E. coli to this hexitol. At least 4 aminoacyl residues in this pro- tein have been suggested to be important in these activities: His-195, His-256, Cys-384, and His-554. Previous evidence has implicated His-554 and Cys-384 as residues that are covalently phosphorylated, in sequence, as intermediates in phosphotransfer to mann- itol. We have constructed a number of site-specific mutants of the mannitol permease at these positions. The properties of proteins in which His-554 or Cys- 384 has been changed are consistent with their essential roles in phosphorylation. We also used these mu- tants to show that intermolecular phosphotransfer between His-554 and Cys-384 can occur in vivo in mem- brane-bound heterodimers consisting of different mutant subunits. The properties of proteins with mutations at position 195 suggest an important role for this residue involving hydrogen bonding, while His- 256 performs no significant function in the mannitol permease. Finally, the phosphorylation and chemore- ception activities for each mutant protein were each roughly in the same proportion to these as described by Stephan and Jacobson (28) but using goat anti-rabbit IgG-horseradish peroxidase conjugate to detect binding of anti-EIImtI (Bio-Rad). Protein concentrations in membrane vesicle preparations were determined by the method of Lowry et al. (29) using bovine serum albumin as the standard. Functional Complementation of Mutant Proteins in Vivo-Mutant genes originally cloned into pGJ9 were subcloned into pBR322 using the SalI and BamHI sites in each vector (12,30). E. coli LGS322 cells were then serially transformed with various painvise combinations of pGJ9 and pBR322 vectors harboring different mutant mtlA genes, selecting for both chloramphenicol (pGJ9) and ampicillin (pBR322) resistance. The resultant cells were then tested for mannitol fermentation on MacConkey mannitol plates and for PEP-dependent phos- phorylation of mannitol as described above. domain role of phosphoex- change, phosphotransfer


Site-specific Mutagenesis of Residues in the Escherichia coli Mannitol Permease That Have Been Suggested to Be Important for Its Phosphorylation and Chemoreception Functions*
(Received for publication, February 28, 1992) Qing-Ping Weng, Jacqueline Elder, and Gary R. Jacobson The Escherichia coli mannitol permease is an integral membrane protein that catalyzes the concomitant transport and phosphorylation of D-mannitol and also acts as the chemoreceptor for chemotaxis of E. coli to this hexitol. At least 4 aminoacyl residues in this protein have been suggested to be important in these activities: His-195, His-256, Cys-384, and His-554. Previous evidence has implicated His-554 and Cys-384 as residues that are covalently phosphorylated, in sequence, as intermediates in phosphotransfer to mannitol. We have constructed a number of site-specific mutants of the mannitol permease at these positions. The properties of proteins in which His-554 or Cys-384 has been changed are consistent with their essential roles in phosphorylation. We also used these mutants to show that intermolecular phosphotransfer between His-554 and Cys-384 can occur in vivo in membrane-bound heterodimers consisting of different mutant subunits. The properties of proteins with mutations at position 195 suggest an important role for this residue involving hydrogen bonding, while His-256 performs no significant function in the mannitol permease. Finally, the phosphorylation and chemoreception activities for each mutant protein were each roughly in the same proportion to these activities in the wild-type protein, showing that these functions of the mannitol permease are tightly coupled under normal physiological conditions. The phosphoenolpyruvate (PEP)'-dependent carbohydrate phosphotransferase system (PTS) couples the transport and phosphorylation of many carbohydrates in a variety of bacterial species (reviewed in Ref. 1). The phosphotransfer reactions involved in this process are detailed below.
Enzyme I (EI) and HPr are general cytoplasmic phosphotransfer proteins of the PTS, while Enzyme I11 (EIII) and Enzyme I1 (EII) are carbohydrate-specific components. For some substrates (e.g. glucose in Escherichia coli), E111 is a soluble, cytoplasmic enzyme that phosphorylates the membrane-bound EII, while for others (e.g. mannitol in E. coli), the functions of E11 and E111 are combined into a single E11 polypeptide with an integral membrane (EII) domain and a cytoplasmic (EIII) domain (2). In addition to participating in PEP-dependent phosphorylation of its substrate, each E11 also catalyzes a phosphoexchange reaction between its phosphorylated product and its substrate, reflective of the phospho-E11 intermediate shown in the sequence of reactions above.
In addition to its carbohydrate transport and phosphorylation functions, the PTS also regulates the activities and expression of non-PTS transport systems (3), and many of the PTS EIIs serve as primary chemotactic receptors for their cognate substrates, at least in the enteric bacteria (4). One of the best characterized of the PTS EIIs is the one that is specific for D-mannitol in E. coli (EIImt' or mannitol permease). This 68-kDa integral membrane protein has been purified (5) and characterized (reviewed in Refs. 6 and 7), and its gene (rntlA ) has been cloned and sequenced (8). It consists of an N-terminal, membrane-bound domain (residues from 1 to about 334) that spans the membrane at least six times (9) and a hydrophilic, cytoplasmic domain (residues from about 335 to 637) that includes an EIII-like subdomain (10,11). The N-terminal domain is responsible for mannitol binding and translocation, while the C-terminal domain carries out the phosphorylation functions of the mannitol permease (12). Recent experiments using biochemical (13), biophysical (14,15), and molecular genetic (12,16) approaches have provided strong evidence that His-554 in the EIII-like subdomain of the mannitol permease is the phosphoacceptor from P-HPr and that this phospho group is subsequently transferred to Cys-384, which then acts as the phosphodonor to mannitol as it enters the cell through the N-terminal domain. In addition to His-554, a t least two other histidyl residues, His-195 and His-256, have been hypothesized to have roles in the activities of the mannitol permease (2,17).
In this report, we have used site-specific mutagenesis to further investigate the roles of His-554, Cys-384, His-195, and His-256 in the various activities of the mannitol permease. Our results characterizing proteins with various mutations at positions 554 and 384 confirm and extend previous conclusions regarding both the roles of the residues found a t these Corp. All enzymic and sequencing reactions were carried out as recommended by the supplier. Goat anti-rabbit IgG-horseradish peroxidase conjugate was purchased from Bio-Rad. Oligonucleotides used for mutagenesis and sequencing were synthesized by Dr. Tolan (Boston University) using a MilliGen (Bedford, MA) model 6500 instrument. A cytoplasmic fraction from Salmonella typhimurium, strain LJ144 (see Table I), was prepared and used as a source for E1 and H P r activities as previously described (21). Other chemicals were reagent grade and were purchased from Sigma.
Bacterial Strains and Plasmids-The bacterial strains and plasmids used are listed in Table I. Plasmid DNA was isolated by the alkaline lysis method (22) and was introduced into the appropriate bacterial strains by the CaC12 transformation procedure of Mandel and Higa (23).
Construction of Mutants-Site-specific mutagenesis of the mtlA gene was performed according to Kunkel (24). The mtlA gene was removed from plasmid pGJ9 (12) using the flanking SalI and BamHI restriction sites and was subcloned into these sites both in phagemid pUC119 yielding pUQWl (see Table   I) and in bacteriophage M13mp19. These were then used to transform E. coli RZ1032 cells (dut-, ung-). Single-stranded DNA was then isolated from these cells and used as a template for mutagenesis. All mutagenic oligonucleotide primers were synthesized such that the region of mismatch was near the center of the primer sequence (see Table 11). Each primer was hybridized with the single-stranded M13 or pUC119 derivative, followed by extension and ligation with the Klenow fragment of DNA polymerase I and T4 DNA ligase. Aliquots of the reaction mixtures were then used to transform competent E. coli MV1190 cells. Phagemids or phage carrying the appropriate mutation were detected either by Southern hybridization using '"P-labeled mutagenic primer (22) followed by DNA sequencing or directly by DNA sequencing using appropriate oligonucleotide primers and the Sequenase version 2.0 protocol (U. S. Biochemical Corp.). A fragment of the mtlA gene containing each mutation in pUQW1 or in M13mp19 (replicative form) was excised using either a SalIISnaBI or a SnaBIIBamHI double digest (10,12), purified by low melt agarose gel electrophoresis, and ligated into the plasmid pGJ9 from which the corresponding portion of the wild-type mtlA gene had been removed. The resulting plasmids were then transformed into E. coli LGS322 (AmtlA (12)) for functional analysis of the mutant protein.
Cell Growth and Preparation of Membrane Vesicles-Bacteria were routinely grown in liquid culture at 37 "C on 2X-YT medium (1.6% tryptone, 1% yeast extract (both from Difco), 1% NaC1) containing 100 pg/ml ampicillin or 30 pg/ml chloramphenicol if necessary for selection. Membrane vesicles were prepared from E. coli LGS322 cells harboring mutant mtlA plasmids as previously described (12).
Assays-Assays of PEP-dependent phosphorylation of D-mannitol and mannito1:mannitol-1-P phosphoexchange activities were performed using either membrane vesicles or permeabilized whole cells as previously described (10). Phosphorylation of EII"" and derived mutant proteins was carried out as follows. ["PIPEP was prepared from [y-"'PIATP by the protocol of Roossien et al. (25) as modified by Stephan et al. (26). Proteins in membrane vesicles were labeled, electrophoresed, and visualized by autoradiography also as described (26).
For a semiquantitative measure of chemotactic ability, E. coli LGS322 cells harboring plasmids containing wild-type or mutant mtlA genes were inoculated (10 p1 of an overnight culture) 1 cm from the edge and just beneath the surface of a soft agar plate of the following composition: 0.25% agar; 10 mM K,HPO,; 1 mM MgC12; 1 mM (NH4)zS04; 1% glycerol; 10 pg/ml each thiamine, histidine, arginine, and methionine, and 30 pg/ml chloramphenicol (pH 7.0). At the same time, 10 pl of a sterile solution of 1% D-mannitol was pipetted just beneath the surface at the center of the plate. Plates were incubated at 30 "C for 24 h. The distance moved by the front of the swarm toward the center of the plate was divided by the average of the two distances moved in the directions normal to this and was called the chemotactic index (a value of 1.0 = no chemotaxis). The expression of wild-type and mutant proteins was estimated in membrane vesicles after electrophoresis on 10% polyacrylamide gels containing SDS (27) by immunoblotting as described by Stephan and Jacobson (28) but using goat anti-rabbit IgG-horseradish peroxidase conjugate to detect binding of anti-EIImtI (Bio-Rad).
Protein concentrations in membrane vesicle preparations were determined by the method of Lowry et al. (29) using bovine serum albumin as the standard.
Functional Complementation of Mutant Proteins in Vivo-Mutant genes originally cloned into pGJ9 were subcloned into pBR322 using the SalI and BamHI sites in each vector (12,30). E. coli LGS322 cells were then serially transformed with various painvise combinations of pGJ9 and pBR322 vectors harboring different mutant mtlA genes, selecting for both chloramphenicol (pGJ9) and ampicillin (pBR322) resistance. The resultant cells were then tested for mannitol fermentation on MacConkey mannitol plates and for PEP-dependent phosphorylation of mannitol as described above. Table 11, site-specific mutants of mtlA were constructed using singlestranded DNA from pUCl19 or M13mp19 into which mtlA had been cloned, and each was confirmed by DNA sequencing. Mutant portions of these genes were subcloned into the plasmid pGJ9 from which the corresponding part of the wildtype mtlA gene had been removed, and each was transformed into E. coli LGS322 (which contains a deletion in the chromosomal mtlA gene). By these procedures, the following mutants in the mannitol permease were obtained2: H554A, H554D, C384H, C384D, H195N, H195R, H195A, and H256A (also see Tables I and 11).

Construction and Expression of Mutant mtlA Genes-As described under "Materials and Methods" and in
To examine whether these mutant proteins were expressed and inserted into the membrane, membrane vesicles were prepared and subjected to Western blot analysis after electrophoresis on SDS gels. As shown in Fig. 1, the mutant proteins were detected at the same apparent molecular mass (65 kDa) and in comparable amounts with the wild-type protein, although some of the mutant proteins (e.g. H195A, H195N, H195R, and H554A) were slightly degraded, presumably by endogenous proteolytic activity.
Phenotypic Properties of Cells Containing EZF" Mutant Proteins-For a qualitative assay of E I P ' activity, E. coli LGS322 cells harboring mutant plasmids were grown on MacConkey mannitol indicator plates. As with wild-type EIImt', colonies producing mutant proteins H256A and H195N were red, indicating efficient transport and metabolism of mannitol. However, colonies producing mutant proteins H195R, H195A, C384H, C384D, H554A, and H554D were white, suggesting that these mutant enzymes are defective in transport and/or phosphorylation of mannitol.
Phosphorylation of EZPt' Mutants--EIImt' is covalently phosphorylated by [32P]PEP in the presence of E1 and HPr (31,32). Further evidence suggested that this protein contains two phosphorylation sites that are intermediates in phosphotransfer to mannitol, His-554 and Cys-384 (13). The phos-Site-specific mutants are denoted by the single letter code of the wild-type residue followed by position number in the primary sequence and the code for the replacement residue. Thus, H554A is the mutant protein in which histidine (H) a t position 554 has been replaced by alanine (A).  CYS-384 + ASP The notation H195A, for example, denotes the resultant mutant protein in which a histidine residue was replaced with an alanine residue a t position 195.
'The deoxyoligonucleotides are written in the 5' to 3' direction. The mutated codons are highlighted in boldface.
photransfer steps were proposed to be the acceptance of a phospho group from phospho-HPr by His-554, the transfer of this group from His-554 to Cys-384, and the transfer from Cys-384 to mannitol (6,7,10,(12)(13)(14)16). If so, mutations in position 554 should completely abolish phosphorylation of EIImt'. As shown in Fig. 2, mutant proteins H554A and H554D were not detectably phosphorylated by [3ZP]PEP. No phosphorylation of these two mutant proteins was observed even at much longer exposures of the same gel used for Fig. 2 (not shown). However, mutants H195A, H195N, H195R, H256A, and C384H were phosphorylated by ["PIPEP, as expected, although some quantitative differences were seen ( Fig. 2; also see "Discussion"). This is consistent with the evidence that His-554 is the phosphoacceptor from phospho-HPr. Effects of Mutations on Mannitol Phosphorylation Activities of EZPt'-Everted membrane vesicles derived from different cells harboring the mutant proteins were used to measure both PEP-dependent mannitol phosphorylation and phosphoexchange activities. The results are presented in Table  111. As expected, replacement of His-554 with either Ala or Asp resulted in undetectable PEP-dependent phosphorylation activity, but phosphoexchange activity was close to that of the wild-type protein. Similar results were obtained recently by van Weeghel et al. (16) for an H554A mutant enzyme. Substitution of His-195 with Asn led to a protein exhibiting nearly 100% of the PEP-dependent activity and 30% of the phosphoexchange activity of the wild-type protein, while substitution of His-256 with Ala had no significant effect on either activity. These results show that neither of these His residues is phosphorylated as an obligatory intermediate in mannitol phosphorylation as had been previously proposed (2,17). However, both H195A and H195R mutant proteins exhibited very low activities in both assays. This could be explained by the possibility that His-195 has a role involving electrostatic interactions; this role could be supplied by His or Asn at this position, but not by Ala or Arg (also cf. "Discussion"). Mutants C384H and C384D were inactive in PEP-dependent phosphorylation of mannitol but still exhibited partial phosphoexchange activity, suggesting that both His and Asp at position 384 can also accept a phospho group from mannitol-1-P. These results are in contrast to the mutant enzyme C384S, which has neither PEP-dependent nor phosphoexchange activity (16).
Zn Vivo Complementation of PEP-dependent Phosphorylation Activity-Much in vitro evidence suggests that functions of E I P ' require an oligomer of the protein, minimally a dimer (reviewed in Refs. 6, 7, and 33). This has been most directly confirmed by demonstration of phospho group transfer between His-554 and Cys-384 on different subunits of EII"" (16,26). T o further investigate this process in vivo, mutant genes encoding C384H, H554A, and H554D proteins were each subcloned from the pGJ9 derivatives (Cm', pACYC184-de- harboring plasmids expressing EII"" a n d its mutants. Membrane proteins (40 pgllane) were separated by SDS-polyacrylamide gel electrophoresis on a 10% gel and probed in Western blots with anti-EII"" antibody as described under "Materials and Methods." Lanes (from left to right): molecular mass standards (from top to bottom: phosphorylase b (97 kDa), bovine serum albumin (68 kDa), ovalbumin (43 kDa), n-chymotrypsinogen (30 kDa), P-lacto-globulin (18 kDa), lysozyme (14 kDa)); mutant H195A (pAQW6); mutant H195N (pAQW2); mutant H195R (pAQW3); mutant H256A (pAQW5); mutant C384H (pAQW7); mutant H554A (pAQW4); mutant H554D (pAQW8); wild-type enzyme (pGJ9); membranes from LGS322 cells with no plasmid. A similar amount of EII"" was found in membranes from cells harboringpAQW9 expressing mutant C384D (not shown on this gel). rived) into the compatible plasmid pBR322 (Amp') bearing a different replication origin. Various combinations of mutant plasmids were then serially transformed into E. coli LGS322 followed by plating on MacConkey mannitol indicator plates containing both chloramphenicol and ampicillin. All of the colonies of cells bearing plasmids encoding C384H and H554A, as well as those bearing plasmids encoding C384H and H554D, were red, in contrast to the colonies with each single mutant gene alone, which were white. This shows that complementation occurs between Cys-384 and His-554 mutant proteins in vivo. To ensure that both plasmids were present in the complemented strains and to rule out that homologous recombination of mutant plasmids had occurred (even though strain LGS322 is recA-), the plasmids from these transformants were prepared and subjected to restriction analysis. These experiments showed that no detectable recombination between plasmids had occurred in the red transformants (data not shown).
To quantitatively estimate the complementation activity between His-554 and Cys-384 mutants, permeabilized cells of the transformants were used for assays of PEP-dependent phosphorylation of mannitol. The results are shown in Fig. 3. The strains containing the genes encoding H554A, H554D, or C384H mutant proteins alone exhibited only background PEP-dependent phosphorylation activity (i.e. that exhibited by strain LGS322 containing no plasmids). However, cells harboring plasmids expressing both C384H and H554A proteins or both C384H and H554D proteins exhibited 25-40% of the PEP-dependent activity of the control strain harboring  pGJ9 (wild-type rntlA). These results provide direct evidence for formation of mutant hetero-oligomers in vivo resulting in intermolecular phosphotransfer from His-554 on one subunit to Cys-384 on another, followed by phosphotransfer to mannitol.
Chemotactic Receptor Activities of EIF" Mutants-The PTS EIIs are the primary chemotactic receptors for their cognate substrates in chemotaxis of E. coli and S. typhimurium to PTS substrates (reviewed in Ref. 4). Although an early report suggested that the transport/phosphorylation and chemotactic receptor functions of EII"" could be at least partially dissected by mutation (34), subsequent attempts to verify this have failed (4). Moreover, there is recent evidence that it is the phosphorylation state of HPr that sends the signal to the Che proteins in taxis to PTS substrates rather than a direct interaction of the receptors (EIIs) with Che proteins as appears to be the case for MCP-mediated taxis (35). If this is true, then it should not be possible to selectively abolish either transport/phosphorylation or chemoreceptor activity in an EII; these activities should be obligatorily coupled. To test this, we measured chemotaxis activities of LGS322 cells harboring plasmids containing wild-type and various mutant mtlA genes using a semiquantitative swarm plate assay (see "Materials and Methods"). Chemotactic responses to mannitol of cells containing various deletion mutants of mtlA or site-specific mutants were proportionally similar to the PEP-dependent mannitol phosphorylation activities of these proteins when compared with the wild-type protein, as shown in Table IV. In contrast, chemotaxis toward aspartate, which is MCP-dependent, was similar in all of these strains. Moreover, several of the mutant proteins tested still bind mannitol normally but were inactive as chemotactic receptors. These results suggest that phosphorylation activity is indeed obligatorily coupled to PTS chemotactic receptor activity, at least in EIImt'.

DISCUSSION
That EIImtl should contain two catalytically important phosphorylation sites was first inferred from the fact that it contains an EIII-like domain (reviewed in Ref. 2). EIIIs are separate proteins for some PTSs (e.g. glucose) and are known in these cases to be phosphorylated on a histidyl residue by phospho-HPr. In the E. coli glucose PTS, E11 specific for glucose is also phosphorylated covalently in a catalytically important fashion (36). For EIImtl, two phosphopeptides have been isolated from the purified protein that had been phosphorylated with PEP, EI, and HPr (13), and it was also demonstrated that PEP-dependent phosphorylation could be genetically dissected from phosphoexchange activity in certain deletion mutants of EIImt' (12). Phosphopeptide analysis (13), and more recently ' "P NMR analysis of purified, phos-phorylated EII"*' (15), provided strong evidence that His-554 and Cys-384 were the sites that were phosphorylated, in sequence, before phosphotransfer to mannitol. That both of these were bona fide intermediates in the PEP-dependent phosphotransfer to mannitol was inferred from an analysis of the stereochemical course of this reaction catalyzed by EI, HPr, and EIImt' (14).
Our results with site-specific mutagenesis of His-554 and Cys-384 are fully consistent with these assignments and confirm and extend recent results of van Weeghel et al. (16) in which these residues were changed to Ala and Ser, respectively. In all of these mutant proteins, PEP-dependent mannitol phosphorylation was abolished. Moreover, of all the mutants described in the present study, only the H554A and H554D proteins were not detectably phosphorylated using [32P]PEP, which would be expected if His-554 were the immediate phosphoacceptor from phospho-HPr.
Also as expected, the C384H mutant was phosphorylated to a lesser extent than the wild-type protein (Fig. 2). Furthermore, our results with proteins in which His-195 or His-256 was mutated rule out that either of these residues could be a catalytically important phosphorylation site as had been previously suggested on the basis of sequence comparisons (2,17). Thus, the H256A mutant protein was fully active in both phosphorylation activities and was phosphorylated to an extent comparable with the wild-type protein, while the H195N protein showed partial phosphorylation activities even though it was phosphorylated to a lesser extent than the wild-type protein, at least in this assay (Table  I11 and Fig. 2). The reasons for the lower phosphorylation levels of all three mutant proteins at position 195 (Fig. 2) remain to be determined.
Interestingly, both C384H and C384D mutant proteins exhibited significant phosphoexchange activity with mannitol-1-P and [14C]mannitol as substrates. These results indicate that both His and Asp at position 384 also can be phosphorylated by mannitol-1-P and that this phospho group can subsequently be transferred to free mannitol. This is in contrast to the C384S protein, which is apparently inactive in phosphoexchange (16). Since both C384H and C384D proteins can still be phosphorylated by phospho-HPr (at least at His-554, Fig. 2), the inactivity of these mutant proteins in PEP-dependent phosphorylation must be due to the inability of a phospho group to be transferred from His-554 to either His or Asp at position 384, either for thermodynamic reasons, stereochemical reasons, or both. If the C384S mutant protein can be phosphorylated by mannitol-1-P, as seems likely on the basis of thermodynamic arguments (phosphoserine in proteins is generally much more stable than phosphohistidine or phosphoaspartate, and in some cases may be more stable than sugar phosphomonoesters (see e.g. Refs. 37 and 38)), its inactivity in phosphoexchange could be explained if phosphotransfer from this intermediate to mannitol does not occur at a detectable rate. The thermodynamics of how mannitol-1-P could phosphorylate Cys, His, or Asp at position 384, however, remain to be determined since this normally would be an endergonic process in each of these cases (37,38). However, it should be noted that phosphoexchange can be measured only at high ratios of mannitol-1-P to mannitol (5,39), as would be predicted from these considerations.
Although our results rule out His-195 as a covalently phosphorylated residue in EIImtl, this residue does appear to play a role in the activities of the protein. While the H195N mutant protein had about 100 and 30% of the activities of the wildtype protein in PEP-dependent and phosphoexchange reactions, respectively, these activities in the H195A and H195R The C-terminal deletion mutants (entries 2-4) are as described (12), and data for PEP-dependent and phosphoexchange activities for these mutants are taken from that reference. For all entries, these two activities are expressed as a percentage of wild-type activity (cf . Table I11 for the saecific activities of the site-saecific mutants).
Mannitol binding was determined as described (12) at a [3H]mannitol concentration of 10 p~ (Q-P. Weng and C. Briggs, unpublished data). In the present study, no attempt was made to determine complete binding iso-therms. Rather, the data are presented qualitatively (+, 250% of the total binding observed for the wild-type protein at this concentration of mannitol, corrected for nonspecific binding).
The chemotactic index (CI) was calculated as described under "Materials and Methods." Since a value of 1.0 indicates no chemotaxis, the relative percentage of chemoreceptor activity (value in parentheses) is equal to ((CI -1.0)/1. mutant proteins were detectable but very low (Table 111). These results could be explained if the hydrogen-bonding ability of the residue at position 195 plays a role, either directly or indirectly, in catalysis. This requirement could be fulfilled by Asn but not by Ala at this position. The inactivity of the H195R mutant is not surprising given the very different chemical properties of imidazole and guanidino groups, and this fact coupled with the fact that the H195N mutant is active rules out that a positive charge at this position is necessary for activity. Further work will be necessary, however, to determine the exact role of His-195 in the activities of EIImtl.
By subcloning mutant mtlA genes from the pACYC184based pGJ9 derivatives into pBR322, we were able to construct strains in a AmtlA background expressing various combinations of proteins containing mutations at positions 384 and 554. In all cases in which an inactive mutant at position 384 was coexpressed with an inactive mutant at position 554, mannitol fermentation in vivo was restored in strain LGS322, and restoration of PEP-dependent mannitol phosphorylation was also observed in permeabilized whole cells of the same strains. These results provide direct evidence for in uiuo hetero-oligomer formation of EIImt' in strains expressing these mutant proteins and extend to the intramembrane in uiuo situation the previous evidence obtained in detergent solution which indicated that intermolecular phosphotransfer can occur between His-554 and Cys-384 (16,26). In fact, it seems likely that this is the major, if not the sole, route for phosphotransfer within the wild-type enzyme, since in vivo complementation activities (Fig. 3) are close to those expected on the basis of the presumed amount of heterodimers relative to inactive homodimers in the membrane (50% of the total) and since dissociated EIImt' has a very low PEP-dependent activity (40). Finally, our results concerning chemotactic activities of cells expressing mutant EIImtl proteins are consistent with the hypothesis that PTS EIIs do not interact directly with the Che signaling proteins (4,35) as do the MCPs, which are the primary receptors for most other chemoattractants (reviewed in Ref. 41). The chemotactic behavior of these cells toward mannitol as a chemoattractant parallels the activities of these mutant proteins in PEP-dependent mannitol phosphorylation but does not correlate with the ability of mutant proteins to The actual three-dimensional arrangement of these helices in the membrane is not known. The C-terminal half consists of two subdomains, each of which contains one phosphorylation site. Functional EIImt' exists at least as dimer (see text). A phospho group from His-554 (PI site) on one subunit is transferred intermolecularly to Cys-384 (P2 site) on the other subunit and finally to incoming mannitol. The interactions between subunits and between domains are not shown. Chemotactic receptor activity of EIImt' is probably due to its modulation of the P-HPr/HPr ratio in the presence of mannitol, which in turn influences the activities of Che proteins. The various domains of EIImt' are not drawn to scale. This figure is similar to Fig.  2 of Ref. 16 but shows the updated model for the topology of the membrane-bound domain (9), the role of Cys-384 in phosphoexchange, the predominant route of phosphotransfer within the protein, and the way in which PTS EIIs are proposed to interact with the Che proteins.
bind mannitol (Table IV). Thus, unlike the MCPs, transport and phosphorylation of its ligand, rather than simply binding, is essential for a PTS E11 to act as a chemoreceptor (at least in the case of EIImtl). Indeed, recent results suggest that it is the phosphorylation state of HPr (which in turn is determined by the combined activities of the PTS EIIs) that communicates, directly or indirectly, with the Che proteins, which themselves are involved in regulating flagellar rotation (35).

CONCLUSIONS
From the results of this and previous work, we present the model shown in Fig. 4 for the various activities of EIImt'. 1) We have confirmed the results of van Weeghel et al. (16) concerning the roles of His-554 (phosphoacceptor from P-HPr) and Cys-384 (phosphodonor to mannitol) in EIImtl. We have also extended these results by showing that mutations at position 554 completely abolish phosphorylation of the protein while mutations at position 384 do not. 2) Previous results (16,26) concerning the ability of the phospho group on His-554 to be transferred intermolecularly to Cys-384 on another subunit of EIPl have been confirmed. We have extended this observation of intermolecular phosphotransfer to the membrane-bound protein in vivo and have provided evidence that this is the predominant, if not the sole, route of phosphotransfer within the enzyme. 3) We have shown that although mutant proteins with His or Asp at position 384 are inactive in PEP-dependent phosphorylation, they still catalyze phosphoexchange (in contrast to the C384S mutant (16)). 4) We have shown that His-256 performs no important function in EII"*l, while His-195 appears to play a role in its activities involving hydrogen bonding. 5) Finally, using these mutants, we have shown that mannitol transport and phosphorylation appear to be essential for EIImt' to act as a chemotactic receptor. These observations are consistent with the hypothesis that it is the HPr/P-HPr ratio that is sensed by the general chemotaxis (Che) proteins during taxis to PTS substrates (35).
In the future, it will be interesting to use site-specific mutagenesis to define further the roles of individual aminoacyl residues in EII"" activities, especially those in the membrane-bound domain that are probably responsible for the translocation event itself.