The intramembrane topography of the mannitol-specific enzyme II of the Escherichia coli phosphotransferase system.

The D-mannitol-specific Enzyme II of the phosphoenolpyruvate-dependent phosphotransferase system of Escherichia coli is an integral cytoplasmic membrane protein responsible for concomitant transport and phosphorylation of this hexitol. We have investigated the intramembrane topography of this enzyme/permease using proteases, membrane-impermeable reagents, and antibodies against the purified protein. The results of these experiments suggest that this protein spans the membrane in a single orientation with a sizeable proportion of its mass extending into the cytoplasm, but with little of the polypeptide exposed at the outside surface of the membrane. Such an orientation is consistent with the reception and transport roles of the mannitol Enzyme II in E. coli.

The D-mannitol-specific Enzyme I1 of the phosphoenolpyruvate-dependent phosphotransferase system of Escherichia coli is an integral cytoplasmic membrane protein responsible for concomitant transport and phosphorylation of this hexitol. We have investigated the intramembrane topography of this enzyme/permease using proteases, membrane-impermeable reagents, and antibodies against the purified protein. The results of these experiments suggest that this protein spans the membrane in a single orientation with a sizeable proportion of its mass extending into the cytoplasm, but with little of the polypeptide exposed at the outside surface of the membrane. Such an orientation is consistent with the reception and transport roles of the mannitol Enzyme I1 in E. coli.
In recent years, it has become clear that most integral membrane transport proteins that have been studied in detail both span the lipid bilayer and are situated in a single orientation with respect to the inside and outside of the system bounded by the membrane. Furthermore, at least two classes of such proteins have been recognized (1,2), those proteins that have most of their mass embedded in the hydrophobic domain of the membrane, and those that have a sizeable fraction of their polypeptide exposed to aqueous environments at either or both faces of the bilayer. An example of the former is the hydrophobic bacteriorhodopsin polypeptide of Halobacterium halobium (3, 4) and examples of the latter include ion-translocating ATPases (5)(6)(7) and other transport proteins such as the anion channel of human erythrocytes (8).
In a previous communication, one of us reported the purification to apparent homogeneity of the protein responsible for the tightly coupled transport and phosphoryation of Dmannitol in Escherichia coli (17). This protein, the mannitol-* This research was supported by a grant from the Boston University Graduate School and supported by National Institutes of Health Grant 1 R01 GM28226 from the National Institute of General Medical at the Annual Meeting of the American Society of Biological Chem-Sciences. Portions of this report were presented in a preliminary form ists, April, 1982 in New Orleans, LA. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed at, Department of Biology, Boston University, 2 Cummington Street, Boston, MA 02215. specific Enzyme 11' of the PTS (for reviews see Refs. 18 and 19), comprises a single kind of polypeptide chain, M, = 60,000, and is integrally associated with the cytoplasmic (inner) membrane of this organism. These conclusions have independently been confirmed recently in E. coli minicells harboring a hybrid plasmid containing the genes necessary for mannitol utilization (20).
We have sought to determine the orientation of Enzyme 11"" in E. coli membranes using limited proteolysis and membrane-impermeable reagents. Our results lead to a working model for the disposition of this protein in the phospholipid bilayer. Since Enzyme IImt' is a multifunctional protein,2 the functions of which can be dissected genetically (21), immunologically (22), and proteolytically (this report), these probes of intramembrane topography should also be useful in elucidating relationships of structure to function in this enzyme.

EXPERIMENTAL PROCEDURES
Materials-Phosphoenolpyruvate (tricyclohexylammonium salt), D-mannitol, D-mannitol-1-phosphate, soybean trypsin inhibitor, bovine pancreatic trypsin inhibitor, PCMPS, and egg white lysozyme (used in spheroplast preparation) were all purchased from Sigma. a-Chymotrypsin (treated with N-a-p-tosyl-L-lysine chloromethylketone) and papain were also from Sigma, while trypsin (treated with ~-l-tosylamide-2-phenylethylchloromethylketone) was a product of Worthington. ~-['~C]Mannitol was purchased from New England Nuclear. IgG fractions from preimmune and anti-Enzyme 11"'" rabbit sera were prepared as previously described (20). Other compounds were obtained from commercial sources and were the highest grade available.
Cell Growth, Enzyme Purification, and Preparation of Spheroplasts-Enzyme 11"" was purified from mid-exponential phase E. coli K12 (strain KL141) by a modified procedure as previously described (22). As a source of the soluble phospho-transfer proteins of the PTS, Enzyme I and HPr, a cytoplasmic fraction from Salmonella typhimurium strain U144 (cpd-401, cysA1150/F'198) was prepared. This strain contains thepts operon on an E. coli episome and overproduces these proteins about 5-fold relative to wild type cells (23). Spheroplasts were formed from early exponential phase E. coli K12, strain KL141, which had been grown on medium 63 (24) supplemented with thiamin (10 pg/ml), arginine (50 pg/ml), uracil (20 pg/ml), and 0.5% mannitol to an ASSO of 0.3. The procedure involved lysozyme-EDTA treatment in the presence of sucrose exactly as described by Osborn et al. (25) for S. typhimurium. Spheroplasts were kept on ice until used and were freshly prepared for each experiment. Only prepara-tions which contained 295% spheroplasts as judged by phase-contrast microscopy were employed in the experiments reported here. When the effects of various reagents or proteases on Enzyme 11"" activity were to be tested on intact spheroplasts, stock solutions of these potential modifiers were dissolved in buffer of the same composition as that in the final spheroplast preparation (3.3 mM Tris-HCI, pH 7.8, containing 0.23 M sucrose and 1 mM EDTA). For some experiments, whole cells or spheroplasts were disrupted to expose the inner face of the cytoplasmic membrane by passage through a French pressure cell at 10, OOO p.s.i. (26).
Modification Experiments-The effects of trypsin, chymotrypsin, papain, anti-Enzyme 11"" antibody, PCMPS, and KMn04 on Enzyme 11"" activity were tested at room temperature on spheroplasts, on French press-ruptured spheroplasts (or whole cells), and on purified Enzyme 11""'. Conditions for each experiment are given in the legends to the corresponding figures or tables. When spheroplasts were used, their integrity was monitored by phase-contrast microscopy before and after each treatment. None of the treatments with these reagents or enzymes caused lysis of >15% of the spheroplasts under the conditions used in these experiments (see also the figure legends).
Phosphorylation and Uptake Assays-Both unfractionated, membrane-bound Enzyme 11"' " and the purified protein in Lubrol PX catalyze the following reactions (17) The size of the sample to be assayed was chosen such that Enzyme 11"'" was present in rate-limiting amounts. After 30 min at 37 "C, mixtures were rapidly cooled to 4 "C and the ["C] mannitol-1-P formed was determined on DEAE-filter paper (Whatman DE81) as previously described (17). Transphosphorylation activity was determined in mixtures containing 0.1 M potassium phosphate, pH 7.0, containing 5 mM MgC12, 10 mM KF, 1 mM dithiothreitol, 1 mM mannitol-1-P, 0.5 p~ ['4C]mannitol (50 pCi/pmol), and the sample containing Enzyme II""' in a total volume of 0.1 ml. They were incubated at 30 "C for 30 min and then diluted with 1 ml of H20 at 4 "C to stop the reaction. ['4C]Mannitol-l-P formed was then determined by the Dowex I-X2 resin column procedure (27). The uptake of [I4C]mannitol into spheroplasts mediated by Enzyme 11'"" was measured after various treatments at room temperature.
["CIMannitol (5 pCi/pmol) was added to treated or untreated spheroplasts to a final concentration of 0.1 mM. At various times, 1-ml aliquots were removed and centrifuged for 1 min in a Eppendorf microcentrifuge. The supernatant was carefully and thoroughly removed, the pellet was resuspended in 1 ml of H20 to lyse the spheroplasts, and this solution was then transferred to a vial containing I0 ml of standard toluene/Triton X-100-based scintillation fluid for counting.

RESULTS AND DISCUSSION
Proteolytic Modification of Enzyme II"""Purified Enzyme 11"' " is rapidly inactivated by trypsin with the concomitant cleavage of its 60,000-dalton polypeptide into fragments approximately half this mass (17,22). In these experiments, the ability of the enzyme to catalyze PEP-dependent phosphorylation of D-mannitOl was measured. We have confirmed and extended these observations, and the results of these experiments are shown in Fig. 1. Both PEPand mannitol-IP-dependent activities of purified Enzyme IImtl in Lubrol PX were inactivated at similar rates by trypsin (0.5 pg/ml). Similarly, chymotrypsin at 0.5 pg/ml inhibited both activities in Ti"€ bin) FIG. 1. Proteolytic inactivation of Enzyme 11"'". Purified Enzyme 11"" (10 pg/ml) in 20 mM Tris-HC1, pH 8.4, containing 1 mM dithiothreitol and 0.5% Lubrol PX was treated with 0.5 pg/ml of either trypsin or chymotrypsin at 23 "C. At the times indicated on the abscissa, samples were removed and a 2-fold molar excess over protease of either soybean trypsin inhibitor (for the trypsin experiments) or bovine pancreatic trypsin inhibitor (for the chymotrypsin experiments) was added. Samples were then assayed for PEP-dependent and transphosphorylation activities as described under "Experimental Procedures." An activity of "100%" corresponds to 30 pmol of mannitol-1-P formed per 30 min/mg of Enzyme 11"" for the PEP-dependent reaction and 0.29 pmo1/30 min/mg for the mannitol-1-P-dependent reaction. "Zero time" activities were independent of whether the enzyme was pretreated with a mixture of protease and inhibitor, or not: M , PEP-dependent activity after trypsin treatment; A-A, mannitol-1-P-dependent activity after trypsin treatment; 0---0, PEP-dependent activity after chymotrypsin treatment; A---A, mannitol-1-P-dependent activity after chymotrypsin treatment. Inset, proteolytic dissection of Enzyme I P ' functions by chymotrypsin in everted membrane vesicles. Everted vesicles prepared from spheroplasts were treated with chymotrypsin (20 pg/ml) at 23 "Cf in spheroplast preparation buffer, and the reaction was stopped at various times as described above: PEP-dependent ( M ) and mannitol-I-P-dependent (0---0) activities after addition of Lubrol PX to a final concentration of 0.5% are shown. a time-dependent manner, although in this case, PEP-dependent phosphorylation was inactivated much more quickly than transphosphorylation. In membranes derived from French press ruptured cells, chymotrypsin also partially dissected these activities (inset, Fig. 1).
In order to determine the localizations of the trypsin-and chymotrypsin-sensitive sites with respect t o the cytoplasmic membrane, we performed the experiments presented in  intact spheroplasts or everted vesicles derived from them were incubated with trypsin (0.6 pg/ml) or chymotrypsin (3 pg/ml) at 23 "C in spheroplast preparation buffer, followed by sampling and addition of inhibitors as detailed in the legend to Fig. 1. Treated samples from spheroplasts were then passed through a French pressure cell (10,000 p.s.i.) to evert the membranes, and all samples were assayed in the PEP-dependent reaction as in Fig. 1: M , spheroplasts after trypsin treatment; A-A, everted vesicles after trypsin treatment; 0---0, spheroplasts after chymotrypsin treatment; A---A, everted vesicles after chymotrypsin treatment. Spheroplast lysis was judged to be 5 (r~_2)% after treatment with chymotrypsin and 12 (&4)% after treatment of spheroplasts with trypsin after 30 min as judged by phase-contrast microscopy. These values correlate well with the slight inactivations seen when intact spheroplasts were treated with these proteases. B, spheroplast suspensions were treated for 30 min at 23 "C in spheroplast preparation buffer with the indicated concentrations of papain before and after lysis in a French pressure cell. Papain was removed by centrifugation with one wash (5,000 X g for 10 min for spheroplasts; 15,000 x g for 30 min for lysed spheroplasts), the pellets were dissolved in spheroplast preparation buffer containing 0.5% Lubrol PX, and they were then assayed for PEP-and mannitol-1-P-dependent activities. Spheroplast lysis was judged to be 510% at the highest concentration of papain used M , PEP-dependent activity of spheroplasts; M , mannitol-1-P-dependent activity of spheroplasts; A---A, PEP-dependent activity of everted vesicles; A---A, mannitol-I-P-dependent activity of everted vesicles. C, spheroplasts (M) and everted vesicles prepared from them (0---0) were both subjected to oxidation with 0.1 m M KMn04 at 23 "C in spheroplast preparation buffer. At various times indicated on the abscissa, samples were removed and dithiothreitol (IO m M final concentration) was added to stop the reaction. They were then assayed directly for PEP-dependent phosphorylation after addition of Lubrol PX to a final concentration of 0.5%. Nearly identical results were obtained when mannitol-1-P-dependent phosphorylation was measured (not shown). Spheroplast lysis was judged to be 55% after 60 min of KMnOs treatment by phase-contrast microscopy.
spheroplasts, and membranes derived from them by French pressure cell lysis, which yields mainly everted ("inside-out") vesicles (26), were then subjected to proteolysis for varying times. Subsequently, these preparations were tested for their abilities to catalyze PEP-dependent phosphorylation of Dmannitol in the presence of the soluble PTS enzymes. The results of these experiments show ( Fig. 2A) that spheroplasts treated with trypsin or chymotrypsin, then lysed in a French pressure cell and assayed, were largely or completely insensitive to inactivation by these proteases. In contrast, everted membrane vesicles treated with trypsin or chymotrypsin rapidly lost their ability to phosphorylate D-mannitOl.
Next, we investigated the effects of increasing concentrations of the relatively nonspecific protease, papain, on Enzyme 11"' " activities in spheroplasts and everted vesicles. Again, nearly complete inactivation of both PEP-dependent and transphosphorylation activities was observed in everted vesicles, while the enzyme in spheroplasts was essentially insensitive even to relatively high concentrations of papain (Fig.   2B). These results demonstrate, therefore, that the sites sensitive to these proteases are exposed to the aqueous environment on the inner, cytoplasmic surface of the inner membrane.
Finally, we tested the effects of these proteases at 5 pg/ml on the transport rate of ['4C]mannitol into spheroplasts (see "Experimental Procedures"). No significant difference in uptake rate (>%) was seen with any protease tested over a 30min period at room temperature (not shown). Therefore, sites on Enzyme 11"" necessary for activity that are cleaved by these proteases do not become available on the outside surface, even during transport of mannitol by the enzyme.

Effects of Antibody on Membrane-bound Enzyme II""-
The preparation of polyclonal antibodies against purified Enzyme IT"" has been described previously (20). These antibodies specifically inhibited the catalytic activities of both purified enzyme and of the protein in crude membrane fractions (22).
In order to determine the localization of these antibody-combining sites with respect to the membrane, we carried out the experiments shown in Fig. 3. Spheroplast suspensions were treated with partially purified preparations of either control (preimmune) or anti-Enzyme 11"' " IgG, and then measured for their abilities to take up ['4C]mannitol. No significant difference in initial uptake rate was observed for the two preparations (Fig. 3). In contrast, when the same spheroplast preparation was lysed in the French press and treated with an identical amount of anti-Enzyme 11"' " IgG, PEP-dependent phosphorylation of mannitol was inhibited more than 90% relative to a control sample (see legend to Fig. 3). Similar results were obtained when the same experiments were performed with "right-side-out" membrane vesicles prepared by osmotic lysis of spheroplasts (19).' Chemical Modification Experiments-Enzyme 11"" is known to be inactivated by reagents that modify free sulfhydryl groups in proteins (17,19 necessary for this reaction was localized on the inside surface of the membrane (28). We have c o n f i i e d and extended these results in our systems. Treatment of whole cells or spheroplasts with 1 mM PCMPS for 10 min at room temperature, followed by removal of excess reagent by centrifugation and washing, had no effect on transphosphorylation activity of Enzyme 11"" which had been extracted by Lubrol P X from membranes prepared from these samples ( Table I). In contrast, if everted vesicles were similarly treated with PCMPS, washed, and extracted, over 90% of Enzyme 11"" transphosphorylation activity was lost relative to a control (Table I).
Finally, we found that PCMPS had no significant effect on mannitol uptake into either whole E. coli cells or spheroplasts (not shown). These results demonstrate that over 90% of the required sulfhydryl groups must be oriented such that they  fraction (0) or an IgG fraction from anti-Enzyme 11"" serum (0). These rates of uptake did not differ significantly from that in the absence of any IgG. In a parallel experiment, aliquots of this same spheroplast preparation were lysed in a French pressure cell and assayed for PEP-dependent phosphorylation of Dmannitol in the presence of amounts of both IgG fractions identical to those used in the uptake experiment. The rate of phosphoryation in the presence of anti-Enzyme 11"" IgG was found to be only 5% of that in the presence of preimmune IgG. TABLE I Activity of Enzyme II"" in whole cells, spheroplasts, and everted vesicles after treatment with para-chloromercuriphenylsulfonate, a membrane-impermeable sulfiydryl reagent Samples were treated with 1 mM PCMPS for IO min at 23 "C in 20 mM Tris-HC1, pH 7.8, (whole cells and everted vesicles) or in 3.3 mM Tris-HCI, pH 7.8, containing 0.23 M sucrose and 1 mM EDTA (spheroplasts). Excess reagent was then removed by centrifugation (5,000 X g for 10 min in the case of spheroplasts and whole cells, and 15,000 X g for 30 min in the case of everted vesicles) and washed, and the pellets were resuspended in the same buffer. Samples were then tested for mannitol-I-P-dependent phosphorylation of D-mannitol (transphosphorylation) after rupturing in a French pressure cell at 10,000 p.s.i. (whole cells and spheroplasts) or directly (everted vesicles) in the presence of Lubrol PX (0.1%) to solubilize Enzyme 11"'". See "Experimental Procedures" for preparation conditions for spheroplasts and everted vesicles. Controls were treated in the same manner as described above without the addition of PCMPS. ~~~~ ~~ are accessible to this reagent only from the inside surface of the membrane. They also show, therefore, that most, if not all, of the Enzyme 11"" molecules must have a single, asymmetric orientation in the cytoplasmic membrane.
In an effort to modify Enzyme 11"" from the outside surface of the cytoplasmic membrane, we investigated the effects of the powerful oxidant, KMn04. This compound has the potential to oxidize nearly all amino acid side chains, and thus might be expected to inactivate many enzymes (29, 30). Because permanganate anion would not be expected to traverse intact membranes rapidly because of its hydrophilicity and charge, we tested the effects of KMn04 on Enzyme 11"" in intact spheroplasts and in everted membrane vesicles. The results of a representative experiment are shown in Fig. 2C.
Rapid time-dependent inactivation of PEP-dependent phosphorylation of D-mannitol occurred with KMn04 when everted membranes were used as the source of Enzyme 11"". In contrast, spheroplasts treated with KMn04 under the same conditions reproducibly lost only part of their PEP-dependent phosphorylating activity, and inactivation reached a plateau by about 30 min of KMn04 treatment (Fig. 2 0 . These results suggest that KMn04 is not highly permeable to the E. coli cytoplasmic membrane, and partially inactivates Enzyme I P ' from the outside surface of the bilayer. Thus, Enzyme IImt' appears to span the phospholipid b i l a~e r .~ Topography of Enzyme II""-The results presented in this report strongly support the schematic model shown in Fig. 4 for the topography of Enzyme 11"" in the cytoplasmic membrane of E. coli. They suggest that Enzyme 11"' " spans the membrane, but that relatively little of the Enzyme 11"" polypeptide is exposed at the outside surface, while a sizeable proportion of the protein most likely protrudes into the cytoplasm. This conclusion is based on the fact that, with the sole exception of KMn04, no reagent or protease tested was able to significantly inactivate Enzyme IFt1 from the outside surface of intact spheroplasts. Our results also show that nearly all (>90%) active Enzyme IImt' molecules have a single, asymmetric orientation in the cytoplasmic membrane of E. coli because the enzyme extracted by Lubrol PX from spheroplasts treated with papain or PCMPS is 290% as active compared with no treatment, while the same extraction of treated everted vesicles yields little active enzyme (cf Table I and Fig. 2B). The orientation of Enzyme 11"" in the membrane shown in Fig. 4 is also consistent with a number of other properties of the protein. Presumably, it must interact with phospho-HPr, and possibly also Enzyme I of the PTS,6 on the cytoplasmic surface of the membrane. This would require that part of the polypeptide be exposed to the cytoplasm. Also, evidence has ' Because KMn04 oxidation reproducibly (3 experiments) inhibited Enzyme 11"" in spheroplasts by about 308, and this inactivation plateaued after about 30 min, we conclude that the partial inactivation seen is not due to diffusion of KMn04 across the membrane and oxidation from the inside. Rather, because 295% of the spheroplasts remained intact throughout the experiment, we believe that KMn04 oxidizes the enzyme from the outside surface, leading to a species with impaired catalytic function relative to the unoxidized enzyme. An alternative explanation, that KMn04 oxidizes phospholipids important for Enzyme 11"' " function, however, cannot be ruled out by these experiments.
'Direct interaction of Enzyme I with membrane-bound Enzymes I1 of the PTS as shown in Fig. 4 has not yet been shown. However, a significant proportion of total cellular Enzyme I (-10%) is bound to membranes after lysis of E. coli, and this Enzyme I is active in the PEP-dependent phosphorylation of PTS substrates, including D-mannitol (G. Jacobson and D. Finlay, unpublished observations). This binding could also be due to indirect association of Enzyme I with an Enzyme I1 via an Enzyme I. HPr . Enzyme I1 complex. pholipid bilayer asymmetrically with relatively little of its mass in contact with the aqueous environment on the outside surface of the cytoplasmic membrane. Sensitivity of the protein to proteases, antibody, PCMPS, and KMn04 from the cytoplasmic face of the membrane suggests that a larger proportion of the polypeptide protrudes at this membrane surface. This is consistent with the necessity of this permease to interact with the cytoplasmic phospho-transfer protein, HPr, and possibly also Enzyme I (En," of the PTS during transport and phosphorylation of D-mannitol. In contrast, the protein need only interact with the small hexitol molecule on the outside surface of the membrane for its transport and chemoreception functions. Enzyme 11'"" has been shown to have only a moderate content of hydrophobic amino acids (44%) which is also consistent with this proposed topography. The quaternary structure of the polypeptide in the membrane has yet to be e~tablished.~ recently been presented that the PTS Enzymes I1 may be covalently phosphorylated as intermediates in sugar phosphorylation (31), and this site must also be exposed on the cytoplasmic surface of the membrane (at least transiently).
Finally, Enzyme 11"'" is not an especially hydrophobic protein (22), which would also be consistent with the conclusion that a considerable portion of its mass is exposed to an aqueous environment in uiuo. This can be contrasted with the protontranslocating protein bacteriorhodopsin, an extremely hydrophobic protein (4), which is mostly embedded in the lipid bilayer of H. halobium (3), and the E. coli lactose permease which is also very hydrophobic (32, 33), and which actively transports its substrate via proton symport rather than by a phosphotransferase type mechanism (33).
It is also interesting that the time-dependent inactivation by chymotrypsin of the PEP-dependent reaction catalyzed either by purified Enzyme IImt' or by the enzyme in membranes is faster than inactivation of mannitol-l-P-dependent 'Various experiments with cross-linking reagents have failed to reveal any distinct oligomeric complex of 60,000-dalton polypeptides either with purified Enzyme 11"" in Lubrol PX, or in minicell membranes in which this protein was specifically labeled (22). Recent experiments, however, suggest that a multimeric complex of the enzyme may be important for its transphosphorylation function (34).
Preliminary experiments show thatpurified Enzyme 11"" is quantitatively converted to a polypeptide with a molecular weight of about 17,000 as determined on dodecylsulfate-polyacrylamide gels under the conditions shown in Fig. 1 (G. Jacobson and D. Kelly, unpublished observation). No other products were detected, although polypeptides with masses below about 10,000 daltons would not have been resolved on the gel system used. In contrast, trypsin converts the purified protein into a 29,000-dalton species under these conditions (22), and does not dissect the two phosphorylating activities (Fig. 1). transphosphorylation (Fig. 1). This result shows that there is at least one site on the enzyme that is more important for PEP-dependent phosphorylation than for transphosphorylation, the structure of which is altered by Chymotryptic hydrolysis.* A reasonable candidate for this site would be a region on the protein whose structural integrity (or conformation) is necessary in the PEP-dependent reaction for interaction with phospho-HPr in the cytoplasm. However, it remains to be determined whether chymotryptic inactivation of either reaction is due to effects on substrate K , values, V,,, of the reaction, or both. Partial dissection of these activities of Enzyme IImt' also recently has been demonstrated immunologically in E. coli (22) and in S. typhimurium by isolation of certain mutants in the gene coding for this protein (21). It will be interesting to identify which domains of the protein are involved in these functions either by determining the sites hydrolyzed by chymotrypsin or by further genetic studies.
The experiments reported here constitute a first step in understanding how the structure of Enzyme 11"' " is related to its in vivo functions of mannitol transport, phosphorylation, and chemoreception. Since Enzyme 11"' " is quantitatively a relative minor protein of the E . coli inner membrane (17), it has not yet been possible to determine the sizes and structures of the membrane-bound and hydrophilic domains, for example by limited proteolysis. The availability of a plasmid carrying the mannitol operon which can be selectively expressed in E. coli minicells (20), however, should allow these questions to be explored in more detail.