A Single Negatively Charged Residue Affects the Orientation of a Membrane Protein in the Inner Membrane of Escherichia coliOnly When It Is Located Adjacent to a Transmembrane Domain*

The orientation of membrane proteins is determined by the asymmetric distribution of charged residues in the sequences flanking the transmembrane domains. For the inner membrane ofEscherichia coli, numerous studies have shown that an excess of positively charged residues defines a cytoplasmic domain of a membrane protein (“positive inside” rule). The role of negatively charged residues in establishing membrane protein topology, however, is not completely understood. To investigate the influence of negatively charged residues on this process in detail, we have constructed a single spanning chimeric receptor fragment comprising the N terminus and first transmembrane domain of the heptahelical G protein-coupled vasopressin V2 receptor and the first cytoplasmic loop of the β2-adrenergic receptor. When fused to alkaline phosphatase (PhoA), the receptor fragment inserted into the inner membrane of E. coli with its N terminus facing the cytoplasm (Nin-Cout orientation), although both membrane-flanking domains had rather similar topogenic determinants. The orientation of the receptor fragment was changed after the introduction of single glutamate residues into the N terminus. Orientation inversion, however, was found to be dependent on the location of the glutamate substitutions, which had to lie within a narrow window up to 6 residues distant from the transmembrane domain. These results demonstrate that a single negatively charged residue can play an active role as a topogenic determinant of membrane proteins in the inner membrane of E. coli, but only if it is located adjacent to a transmembrane domain.

The orientation of membrane proteins is determined by the asymmetric distribution of charged residues in the sequences flanking the transmembrane domains. For the inner membrane of Escherichia coli, numerous studies have shown that an excess of positively charged residues defines a cytoplasmic domain of a membrane protein ("positive inside" rule). The role of negatively charged residues in establishing membrane protein topology, however, is not completely understood. To investigate the influence of negatively charged residues on this process in detail, we have constructed a single spanning chimeric receptor fragment comprising the N terminus and first transmembrane domain of the heptahelical G protein-coupled vasopressin V 2 receptor and the first cytoplasmic loop of the ␤ 2 -adrenergic receptor. When fused to alkaline phosphatase (PhoA), the receptor fragment inserted into the inner membrane of E. coli with its N terminus facing the cytoplasm (N in -C out orientation), although both membrane-flanking domains had rather similar topogenic determinants. The orientation of the receptor fragment was changed after the introduction of single glutamate residues into the N terminus. Orientation inversion, however, was found to be dependent on the location of the glutamate substitutions, which had to lie within a narrow window up to 6 residues distant from the transmembrane domain. These results demonstrate that a single negatively charged residue can play an active role as a topogenic determinant of membrane proteins in the inner membrane of E. coli, but only if it is located adjacent to a transmembrane domain.
In the past decade, two related hypotheses have been proposed to define the structural features of membrane proteins that determine the establishment of their correct orientation (topology) in the bilayer. For proteins in the inner membrane of Escherichia coli, the "positive inside" rule (1) predicts that loops with an excess of positively charged residues are directed to the cytoplasmic face of the membrane, and many studies have demonstrated that positively charged residues indeed inhibit translocation (for review see Ref. 2). For membrane proteins in the eucaryotic endoplasmic reticulum membrane, the related "charge difference" rule postulates that the charge difference between the flanking segments determines orienta-tion and that negatively and positively charged residues contribute equally (3). The charge difference rule was recently confirmed in yeast with fragments of the pheromone receptor Ste2p (4,5). Folding of an N-terminal domain of a membrane protein impairs translocation and can also influence the orientation (6).
For the inner membrane of E. coli, it was demonstrated that the positive inside rule is based on the electrochemical membrane potential (positive and acidic outside) (7,8). The membrane potential impairs the translocation of positively charged residues and facilitates that of negatively charged residues (7,8). Although the strong topogenic potential of positively charged residues for such an electrophoretic mechanism has been consistently demonstrated, the significance of negatively charged residues is not so clear and has been reported to occur only under certain conditions. From experiments with the double spanning (N out -C in ) leader peptidase (Lep) of E. coli, these were defined as the presence of very high numbers of negatively charged residues (3-4 Asp or Glu residues) (9) and the position-specific attenuation of positively charged residues lying in conformational vicinity to them (10). Experiments with a fusion protein consisting of the N-terminal tail of the phage Pf3 coat protein and Lep indicated that a decreased hydrophobicity of the corresponding transmembrane domain might be another requirement since, in this case, the influence of the membrane potential and hence that of negative charges increases (11). In contrast to those effects that are detectable only under certain conditions, an active and direct role of negatively charged residues as topogenic determinants in the inner membrane of E. coli was also reported using the single spanning (N out -C in ) wild-type Pf3 protein (12). Negatively charged residues were found essential to promote the translocation of the N terminus of this protein and thus to establish the correct topology, even in the absence of positively charged residues (12).
Here we have investigated whether negatively charged residues can play an active role as topogenic determinants for proteins in the inner membrane of E. coli and whether there might be a critical distance from the transmembrane domain within which negatively charged residues are effective. As a model, we used a single spanning chimeric receptor fragment comprising the N terminus and TM1 1 of the heptahelical G protein-coupled vasopressin V 2 receptor and the IL1 domain of the ␤ 2 -adrenergic receptor. This fragment inserted with an N in -C out orientation and was thus suited for assessment of the translocation of the N terminus in the presence of additional negatively charged residues. To determine the orientation of this protein in vivo, the PhoA protein of E. coli (13) was fused C-terminally. PhoA is only active when translocated into the periplasm and is thus a suitable marker for membrane protein orientation (13). ␤-Galactosidase (LacZ) fusions, which are active only when the LacZ moiety is located in the cytoplasm (14), were used as controls. We show that even a single negatively charged residue can play an active role as a topogenic determinant. In addition, we have defined for the first time a window region and show that single negatively charged residues are only effective within a stretch of 6 residues from the transmembrane domain.
DNA Manipulations-Standard DNA preparations and manipulations were carried out. For site-directed mutagenesis, the QuikChange TM site-directed mutagenesis kit (Stratagene) was used without subcloning procedures. The nucleotide sequences of DNA fragments were verified using the FS dye terminator kit from Perkin-Elmer (Weiterstadt, Germany).
Construction of V 2 , ␤ 2 , Chimeric, and Mutant Receptor Fragments-A schematic depiction of the PhoA and LacZ fusion proteins used in this study is shown in Fig. 1. Fusions of PhoA and LacZ to a V 2 receptor fragment comprising the N terminus, TM1, and IL1 (71 amino acids) in the E. coli expression vector pTRC99A have been described previously (pPROV71.PhoA and pPROV71.LacZ) (16).
For the construction of an equivalent fragment of the ␤ 2 receptor, the corresponding cDNA fragment was amplified by polymerase chain reaction from the original sequence (5Ј-primer, 5Ј-CCCCAGCCAGTAAG-CTTACCTGCCAGACTG-3Ј; and 3Ј-primer, 5Ј-GAAGTAGTTGGGATC-CGTCTGCAGACGCTCGAA-3Ј). The 3Ј-primer introduced a novel BamHI site downstream of the stop codon. The polymerase chain reaction fragment was cloned into pTRC99A using the internal NcoI site of the ␤ 2 receptor cDNA, which overlaps the start codon, and the newly inserted BamHI site. The phoA gene of pPROV71.PhoA (see above) was inserted as a BamHI/HindIII fragment into the resulting plasmid, and the lacZ gene of pPROV71.LacZ (see above) was inserted as a BamHI fragment. The resulting plasmids were designated pPRO␤66.PhoA and pPRO␤66.LacZ and encode PhoA and LacZ fusions, respectively, to a ␤ 2 receptor fragment comprising the N terminus, TM1, and IL1 (66 amino acids) (see Fig. 1).
For the construction of chimeric receptor fragments, BstZ17I sites were introduced by site-directed mutagenesis into plasmids pPROV71.PhoA and pPRO␤66.PhoA in the sequences encoding the TM1/IL1 interfaces. The primer sequences were 5Ј-GGTGCTGGCGG-CCGTATACCGGCGGGGCC-3Ј (and its complementary equivalent) for pPROV71.PhoA and 5Ј-GGTCATCACAGCCGTATACAAGTTCGAGC-G-3Ј (and its complementary equivalent) for pPRO␤66.PhoA. The BstZ17I/XbaI fragments of the resulting plasmids were exchanged reciprocally. The resulting plasmid pV 2 /IL␤ 2 .PhoA encodes a PhoA fusion to a receptor fragment comprising the N terminus and TM1 of the V 2 receptor and IL1 of the ␤ 2 receptor. The resulting plasmid p␤ 2 / ILV 2 .PhoA encodes a PhoA fusion to a receptor fragment comprising the N terminus and TM1 of the ␤ 2 receptor and IL1 of the V 2 receptor (see Fig. 1). For the construction of the corresponding LacZ fusion, the BamHI/XbaI fragments of plasmids pV 2 /IL␤ 2 .PhoA and p␤ 2 /ILV 2 .PhoA encoding the PhoA portions were replaced by that of pPROV71.LacZ (see above) encoding the LacZ portion. The resulting plasmids were designated pV 2 /IL␤ 2 .LacZ and p␤ 2 /ILV 2 .LacZ, respectively.
For the construction of charge mutants within the N terminus or the IL domain of the V 2 /IL␤ 2 .PhoA receptor fragment, the corresponding plasmid was used directly for site-directed mutagenesis. 2 For construction of the corresponding LacZ fusions, the BamHI/XbaI fragments of the resulting mutant plasmids encoding the PhoA portions were replaced by that of pPROV71.LacZ (see above) encoding the LacZ portion.
Growth Conditions and PhoA and LacZ Activity Assays-E. coli CC118 cells were cultivated in LB medium (10 g/liter Bacto-peptone, 5 g/liter yeast extract, and 5 g/liter NaCl) or on LB agar plates at 37°C. Ampicillin was used at a final concentration of 100 g/ml. For the determination of whole cell PhoA activity, 20 ml of LB medium were inoculated with a 400-l overnight culture of an E. coli CC118 clone and grown at 37°C for 2 h with aeration. Cells were induced with 1 mM IPTG and grown for an additional 45 min. The PhoA and LacZ activity assays were based on the protocols of Brickman and Beckwith (17) and Miller (18), respectively.
To determine PhoA activity, bacteria from a 100-l cell suspension were harvested (10 min, 8000 ϫ g, 4°C), and the supernatant was removed. The pellet was supplemented with 300 l of 1 M Tris-HCl (pH 8.0), 25 l of 0.1% SDS, and 25 l of chloroform. Samples were stirred for 20 min at 28°C. After phase separation, 200 l of the aqueous upper phase were transferred into a microtiter plate well. Reactions were started by the addition of 25 l of a solution containing 1 M Tris-HCl, 5 mg/ml p-nitrophenyl phosphate, and 5 mM MgCl 2 (pH 8.0). Reactions were incubated for 170 min at 28°C, and A 415 was measured on a microtiter plate reader.
To determine LacZ activity, bacteria from a 100-l cell suspension were harvested and supplemented with 300 l of buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , and 50 mM ␤-mercaptoethanol (pH 7.0)), 25 l of 0.1% SDS, and 25 l of chloroform. Samples were stirred for 20 min at 28°C. After phase separation, 200 l of the aqueous upper phase were transferred into a microtiter plate well. Reactions were started by the addition of 25 l of a solution containing 100 mM K 2 HPO 4 and 4 mg/ml o-nitrophenyl-␤-D-galactoside (pH 7.0). Reactions were incubated for 170 min at 28°C, and A 415 was measured on a microtiter plate reader.
For the detection of PhoA fusions, filters were blocked for 1 h with blocking buffer (10 mM Tris-HCl, 0.9% NaCl, 1% casein, and 1% gelatin (pH 7.2)), supplemented with anti-PhoA antibodies (1:4000 dilution), and incubated for 2 h at room temperature. Filters were washed three times (10 min each) with wash buffer (10 mM Tris-HCl and 0.9% NaCl (pH 7.2)), and anti-rabbit 125 I-IgG was added to a final concentration of 1 g/ml (1 Ci/ml). Filters were incubated for 2 h at room temperature, washed three times (5 min each) with wash buffer, dried, and exposed to x-ray film (1 day).
For the detection of LacZ fusions, filters were blocked for 1 h with buffer A (20 mM Tris-HCl, 150 mM NaCl, 5% low fat milk powder, and 1% Triton X-100 (pH 7.0)) and incubated with monoclonal anti-LacZ antibodies (1:4000 dilution in buffer A) for 1 h at room temperature. Filters were washed four times (15 min each) with buffer A and incubated with alkaline phosphatase-conjugated anti-rabbit IgG (1:5000 dilution in buffer A) for 1 h at room temperature. Filters were washed four times (10 min each) with buffer A; twice (10 min each) with 20 mM Tris-HCl, 150 mM NaCl, and 1% Triton X-100, (pH 7.0); and once (5 min) with 10 mM Tris-HCl, pH 9.5. Filters were incubated in staining solution (0.56 mM 5-bromo-4-chloro-3-indolyl phosphate and 0.48 mM nitro blue tetrazolium) until bands became visible.
Fractionation of E. coli CC118 Clones and Trypsin Digestion of PhoA Fusion Proteins-The protease sensitivity of solubilized PhoA fusion proteins was used to localize the PhoA moieties and to control the PhoA activity data. Prior to membrane protein solubilization, spheroplasts were prepared according to the method of Koshland and Botstein (20) to facilitate cell lysis. 20 ml of LB medium containing 100 g/ml ampicillin were inoculated with a 400-l overnight culture of an E. coli CC118 clone and grown at 37°C for 2 h with aeration. Cells were induced with 1 mM IPTG and grown for 45 min at 37°C with aeration. Bacteria were harvested (10 min, 8000 ϫ g, 4°C) and resuspended in 900 l of ice-cold osmotic shock buffer (100 mM Tris-HCl, 0.5 mM EDTA, and 500 mM sucrose (pH 8.0)). After the addition of 100 l of lysozyme solution (2 mg/ml shock buffer) and 900 l of ice-cold H 2 O, cells were incubated for 5 min on ice to allow the formation of spheroplasts. The suspension was supplemented with 80 l of 500 mM MgCl 2 , and spheroplasts were collected by centrifugation at 1500 ϫ g for 15 min at 4°C. After removal of the supernatant containing periplasmic proteins, spheroplasts were broken in 1 ml of lysis buffer (1% Triton X-100, 0.1% SDS, 50 mM Tris-HCl, 150 mM NaCl, and 1 mM Na-EDTA, (pH 8.0)), and membrane proteins were solubilized. The solution was divided into two 500-l aliquots. One aliquot was incubated with trypsin (1 g/l final concentration) for 30 min at 4°C on an Eppendorf shaker. Protease digestion was stopped with 50 l of Complete protease inhibitor mixture. The control sample was supplemented immediately after membrane protein solubilization with protease inhibitor. For the detection of the PhoA fusions by immunoblot analysis, 20-g protein samples were applied per lane.

RESULTS
Construction and Orientation of V 2 , ␤ 2 , and Chimeric Receptor Fragments-We have previously shown that a V 2 receptor fragment comprising the N terminus, TM1, and IL1 ( Fig. 1) is inserted with its correct N out -C in orientation in the inner membrane of E. coli (16). In contrast, Lacatena et al. (21) demonstrated that the corresponding fragment of the ␤ 2 receptor adopts an inverted N in -C out orientation. Both results are consistent with the positive inside rule; IL1 of the V 2 receptor contains significantly more positively charged residues than IL1 of the ␤ 2 receptor (16,21). To investigate the process of topology determination with high sensitivity, it is advantageous to use a model protein with similar topogenic determinants on either side of the transmembrane domain. In this case, orientation may be influenced by the addition of single charged residues. With this in mind, we constructed a ␤ 2 receptor fragment comprising the N terminus, TM1, and IL1 (PRO␤66) (Fig. 1), corresponding to the previously described V 2 receptor fragment PROV71 (Fig. 1) (16). We next generated chimeric receptor fragments by exchanging the IL1 domains reciprocally. The resulting receptor fragment ␤ 2 /ILV 2 ( Fig. 1) contains the N terminus and TM1 of the ␤ 2 receptor and the IL1 domain of the V 2 receptor, whereas V 2 /IL␤ 2 (Fig. 1) contains the N terminus and TM1 of the V 2 receptor and the IL1 domain of the ␤ 2 receptor. To determine the orientation of the receptor fragments in vivo, they were cloned into the E. coli expression vector pTRC99A, and PhoA and LacZ enzyme moieties were fused C-terminally. The PhoA-and LacZ-negative E. coli strain CC118 was transformed with the expression plasmids. Cells were induced with IPTG, and PhoA and LacZ ac-tivity assays were performed ( Fig. 2A).
As described previously, high LacZ activity, but no PhoA activity, was found in cells expressing the V 2 receptor fragment PROV71, demonstrating its correct N out -C in orientation (16). In contrast, and in agreement with the data of Lacatena et al. (21), high PhoA activity, but no LacZ activity, was detected in cells expressing the ␤ 2 receptor fragment PRO␤66, demonstrating its inverted N in -C out orientation. When the IL1 domain of the ␤ 2 receptor fragment is replaced by that of the V 2 receptor (chimera ␤ 2 /ILV 2 ), the loss of PhoA activity and the acquisition of LacZ activity demonstrate that the correct N out -C in orientation is established. These results are consistent with the positive inside rule. The number of positively charged residues in IL1 is increased from 2 residues in the ␤ 2 receptor fragment to 5 residues in the chimeric fragment. If the IL1 domain of the V 2 receptor fragment is replaced by that of the ␤ 2 receptor (chimera V 2 /IL␤ 2 ), the correct N out -C in orientation of the PhoA fusion protein is changed to the N in -C out orientation as demonstrated by the appearance of high PhoA activity. (No proteolytic degradation products indicative of residual N out -C in -oriented fragments were observed, indicating that this orientation switch is complete (see below).) The topogenic determinants of the two membrane-flanking domains of the V 2 /IL␤ 2 .PhoA fusion protein, however, seem to be rather similar. This is indicated by the corresponding LacZ fusion protein, which also displayed an active phenotype, demonstrating that at least part of the V 2 /IL␤ 2 .LacZ fusion protein remains in the correct N out -C in orientation. (Although not demonstrated directly, the virtual halving of the LacZ activity compared with that of the PROV71.LacZ receptor fragment suggests that this fusion protein adopts both orientations.) To demonstrate that the observed differences in the PhoA or LacZ activities of the fusions are not caused by different expression levels, the fusion proteins were also quantified in whole cell lysates by immunoblot analyses using polyclonal anti-PhoA and monoclonal anti-LacZ antibodies (Fig. 2B). For all PhoA fusion proteins, two specifically stained bands with apparent molecular masses of 55 and 50 kDa were detected. The observed molecular mass of the upper 55-kDa band is in good agreement with the calculated sizes of the receptor fragments plus their PhoA moieties (54.9 kDa, PROV71.PhoA; 54.5 kDa, PRO␤66.PhoA; 54.8 kDa, V 2 / IL␤ 2 .PhoA; and 54.7 kDa, ␤ 2 /ILV 2 .PhoA). This band thus seems to represent the intact fusion proteins. The lower 50-kDa band most likely represents proteolytic degradation products. Proteolysis of PhoA fusion proteins was also observed in pre-FIG. 1. Construction of V 2 , ␤ 2 , and chimeric receptor fragments. The amino acid sequences of the receptor fragments comprising the N terminus, TM1, and IL1 domains are shown. The lengths of the putative transmembrane domains were derived from hydrophobicity plots. PROV71 represents the V 2 receptor fragment, and PRO␤66 the ␤ 2 receptor fragment. The chimeric fragments ␤ 2 /ILV 2 and V 2 /IL␤ 2 were constructed by exchanging the IL1 domains reciprocally. PhoA or LacZ was fused C-terminally to all receptor fragments. Positively charged residues are indicated by closed circles, and negatively charged residues by open circles.
vious studies (e.g. Refs. 15 and 22). It is not known whether this degradation occurs in vivo or during cell lysis. In any case, the immunoblot demonstrates that all PhoA-tagged receptor fragments are expressed in roughly similar amounts and that the observed PhoA activities are not caused primarily by different expression levels. For each of the corresponding LacZ fusion proteins, a specifically stained band with an apparent molecular mass of 125 kDa and a double band of 90 and 85 kDa were detected (Fig. 2B). The observed molecular mass of the upper band is in good agreement with the calculated sizes of the receptor fragments plus their LacZ moieties (123 kDa, PROV71.LacZ; 122.6 kDa, PRO␤66.LacZ; 122.9 kDa, V2/ IL␤ 2 .LacZ; and 122.8 kDa, ␤ 2 /ILV 2 .LacZ). This band thus seems to represent the intact fusion proteins. The lower band most likely represents degradation products. Similar to the PhoA fusions, the immunoblot demonstrates that all LacZ fusions are expressed in roughly similar amounts and that the observed activities are not caused by different expression levels.
For the PhoA fusions, the orientations were also verified by protease sensitivity assays. Mature PhoA protein is proteaseresistant, but only after translocation into the oxidizing environment of the periplasm, where an intramolecular disulfide bond is formed (23). In contrast, it is protease-sensitive in the reducing environment of the cytoplasm, where the critical disulfide bond cannot be formed. Upon protease treatment, a solubilized PhoA fusion protein should be converted to the size of mature protease-resistant PhoA if the PhoA portion reaches the periplasm. In contrast, it should be completely digested when the PhoA portion is located in the cytoplasm. To assess for the localization of the PhoA moieties of the receptor fragments, the E. coli CC118 clones expressing the fusion proteins were converted to spheroplasts to facilitate subsequent cell disruption. The spheroplasts were lysed in the presence of SDS and Triton X-100, and solubilized membrane proteins were treated with trypsin. The PhoA fusions were detected by immunoblotting with polyclonal anti-PhoA antibodies (Fig. 3). For the PRO␤66.PhoA and V 2 /IL␤ 2 .PhoA receptor fragments, the two specific 55-and 50-kDa bands were detected in the untreated controls as described above for the total cell lysates (Fig. 2B). Upon trypsin treatment, both bands were shifted to a single band with an apparent molecular mass of 48 kDa, corresponding exactly to the calculated size of mature proteaseresistant PhoA. The protease resistance of the PhoA moieties of these receptor fragments demonstrates their N in -C out orientation. In contrast, the PROV71.PhoA and ␤ 2 /ILV 2 .PhoA receptor fragments were completely digested upon trypsin treatment, demonstrating the protease sensitivity of their PhoA moieties and thus their complete N out -C in orientation. In the untreated controls of these two receptor fragments, the intact 55-kDa bands were accompanied by several degradation products, which were not observed in total cell lysates. Spheroplast formation may activate additional endogenous E. coli proteases that further degrade the protease-sensitive PhoA moieties of these two receptor fragments. The fact that these additional degradation products were not detectable in the untreated lanes of the chimeric V 2 /IL␤ 2 .PhoA receptor fragment supports the conclusion that this receptor fragment inserts completely in the N in -C out orientation. In summary, the results of the protease sensitivity assays are entirely consistent with those of the PhoA activity assays.
Negatively Charged Residues Efficiently Invert the Orientation of V 2 /IL␤ 2 .PhoA-To assess the precise influence of negatively charged residues on protein topology determination in the inner membrane of E. coli, we used the PhoA-tagged chimeric receptor fragment V 2 /IL␤ 2 .PhoA. It is inserted with an N in -C out orientation, but seems nevertheless to have rather similar topogenic determinants in its membrane-flanking sequences (see above) and should thus represent a good model for these studies. To investigate whether negatively charged residues may play an active topogenic role, we introduced clustered (1-3) negatively charged glutamate residues into the N terminus of this fragment (Fig. 4A). E. coli CC118 cells were transformed with the respective expression plasmids, and the orientation of the fusion proteins was assessed by PhoA activity assays (Fig. 4B). In the proximity of the transmembrane domain, 3 glutamate residues (P34E/L35E/L36E), 2 glutamate residues (L35E/L36E), and even a single glutamate residue (L36E) were sufficient to establish the opposite N out -C in orientation as demonstrated by the complete loss of PhoA activity. A more distantly introduced single glutamate residue (20 residues from the transmembrane domain, S21E) had only a barely detectable effect. More glutamate residues may be needed in this region to invert the orientation since 2 (P20E/S21E) or 3 (L19E/P20E/S21E) glutamate residues decreased the PhoA activity further. All mutant V 2 /IL␤ 2 .PhoA receptor fragments were detected in similar amounts by immunoblot analysis (Fig.   4C), demonstrating that the observed PhoA activities are not caused by different expression levels. The activities of clones expressing the corresponding LacZ fusions were used as controls for the PhoA activity data (Table I). A strong increase in LacZ activity, indicating stabilization of the N out -C in orientation, was observed when the 3 glutamate residues (P34E/L35E/ L36E), the 2 glutamate residues (L35E/L36E), or the single glutamate residue (L36E) was introduced in the proximity of the transmembrane domain. The more distantly introduced glutamate residue (S21E) had again only a minor effect, which appeared to increase after progressive addition of further glutamate residues to this region (P20E/S21E and L19E/P20E/ S21E). The data obtained with LacZ fusions thus strongly support those obtained with PhoA fusions.
A Single Negatively Charged Residue Affects the Orientation of V 2 /IL␤ 2 .PhoA Only if It Is Located within a Narrow Window of 6 Residues from the Transmembrane Domain-Our results imply that there might be a critical distance within which a single glutamate residue is effective. To address this question, we introduced such residues at a distance of 2-7, 10, 13, and 20 residues from the transmembrane domain into the N terminus FIG. 4. Effect of individual and clustered negatively charged residues on the orientation of the chimeric V 2 /IL␤ 2 .PhoA receptor fragment. A, depiction of the constructs used. The residues in the N terminus that were exchanged for glutamate residues are circled. The arrows indicate a reversal of the orientation. B, PhoA activity of E. coli CC118 clones expressing the constructs depicted in A. Activity assays were performed with total lysates from 9.5 ϫ 10 7 cells. Each bar shows the PhoA activity in arbitrary units (V 2 /IL␤ 2 .PhoA without additional negatively charged residues ϭ 1) and represents the means Ϯ S.D. from three individual experiments, each performed in quadruplicate. C, SDS-PAGE/immunoblot analysis of the total lysates shown in B. Fusion proteins of 3 ϫ 10 6 cells were probed with rabbit polyclonal anti-PhoA antibodies and anti-rabbit 125 I-IgG. The immunoblot is representative of three independent experiments. Vector-transformed E. coli CC118 cells were used as a control. WT, wild-type. a LacZ activities are shown in arbitrary units (V 2 /IL␤ 2 PhoA without additional negatively charged residues ϭ 1) and represent the mean Ϯ S.D. from three individual experiments, each performed in quadruplicate. LacZ activity assays were performed with total lysates from 1.9 ϫ 10 8 cells. Mature PhoA is protease-resistant when a critical intramolecular disulfide bond is formed in the periplasm, but protease-sensitive when retained in the cytoplasm, where disulfide bonding does not occur. To assess for protease sensitivity, E. coli CC118 clones expressing the PROV71.PhoA, PRO␤66. PhoA, ␤ 2 /ILV 2 .PhoA, and V 2 /IL␤ 2 .PhoA receptor fragments were converted to spheroplasts and lysed in the presence of SDS and Triton X-100. Solubilized proteins were treated with trypsin (ϩ) or left untreated (Ϫ). Immunoreactive proteins were detected with rabbit polyclonal anti-PhoA antibodies and anti-rabbit 125 I-IgG. Vector-transformed E. coli CC118 cells were used as a control. The immunoblot is representative of three independent experiments. of the chimeric V 2 /IL␤ 2 .PhoA receptor fragment construct (Fig.  5A). E. coli CC118 cells were transformed with the expression plasmids, and the orientation of the fusion proteins was assessed by PhoA activity assays (Fig. 5B) and immunoblotting (Fig. 5C). A strong decrease in PhoA activity, demonstrating the establishment of the N out -C in orientation, was observed only for those glutamate substitutions lying within 6 residues of the transmembrane domain (A39E, R38E, A37E, L36E, and L35E). The more distant glutamate substitutions (P34E, T31E, P28E, and S21E) had only small or negligible effects. All mutant receptor fragments were expressed in similar amounts as demonstrated by immunoblotting of total cell lysates (Fig. 5C). The different PhoA activities are thus not caused by different expression levels. The PhoA activity data are strongly supported by those for the corresponding LacZ fusions (Table I).
Here, too, a major increase in LacZ activity, indicating stabilization of the N out -C in orientation, was observed only for those glutamate substitutions that lie within the 6-residue window of the transmembrane domain (A39E, R38E, A37E, L36E, and L35E). The more distantly introduced glutamate residues (P34E, T31E, P28E, and S21E) had minor effects. In summary, our results demonstrate that even a single glutamate residue can play an active role as a topogenic determinant, but only when it is present in the proximity of the transmembrane domain.

Positively Charged Residues in the IL1 Domain of V 2 / IL␤ 2 .PhoA Affect Orientation Only When Present in Larger
Numbers-We finally assessed whether positively charged residues might also serve as topogenic determinants in our experimental system. Single and clustered (1-3) arginine residues were introduced into the IL1 domain of the chimeric V 2 / IL␤ 2 .PhoA receptor fragment (Fig. 6A), and the orientation of the fusion proteins was assessed by PhoA activity assays (Fig.  6B) and immunoblotting (Fig. 6C). Consistent with the positive inside rule, a complete loss of PhoA activity demonstrating the establishment of the N out -C in orientation was observed when 2 (Y63R/F65R) or 3 (Y63R/F65R/L68R) positively charged residues were present. Individual positively charged residues (Y63R, F65R, L68R, and T70R), however, had only minor or negligible effects. The immunoblot with total cell lysates demonstrates that all mutant receptor fragments were expressed in similar amounts (Fig. 6C). For the corresponding LacZ fusions, complementary activity data were obtained (Table I), i.e. a strong activity increase occurred after the introduction of 2 (Y63R/F65R) or 3 (Y63R/F65R/L68R) positively charged residues, but only minor effects were observed after the introduction of single charges (Y63R, F65R, L68R, and T70R). DISCUSSION We have shown that the introduction of even a single glutamate residue into the N terminus of V 2 /IL␤ 2 .PhoA can invert the orientation of this receptor fragment. In contrast, 2 or more additional arginine residues are needed in the IL1 domain to FIG. 5. Delineation of a window region for the effect of individual negatively charged residues on the orientation of the chimeric V 2 /IL␤ 2 .PhoA receptor fragment. A, depiction of the constructs used. The residues in the N terminus that were exchanged for glutamate residues are circled. The arrows indicate a reversal of the orientation. B, PhoA activity of E. coli CC118 clones expressing the constructs depicted in A. Activity assays were performed with total lysates from 9.5 ϫ 10 7 cells. Each bar shows the PhoA activity in arbitrary units (V 2 /IL␤ 2 .PhoA without additional negatively charged residues ϭ 1) and represents the means Ϯ S.D. from three individual experiments, each performed in quadruplicate. C, SDS-PAGE/immunoblot analysis of the total lysates shown in B. Fusion proteins of 3 ϫ 10 6 cells were probed with rabbit polyclonal anti-PhoA antibodies and antirabbit 125 I-IgG. The immunoblot is representative of three independent experiments. Vector-transformed E. coli CC118 cells were used as a control. WT, wild-type. cause an equivalent effect. The weaker influence of positively charged residues is not surprising if one assumes that the IL1 domain (see Fig. 1) and the attached 452-residue PhoA moiety are transported via the Sec machinery over the inner membrane, which seems to be the general mechanism for all translocated fragments longer than 60 residues (24). In Sec-dependent transport, translocation of positively charged residues is facilitated (25), thus clarifying their weaker influence. In contrast to the IL1/PhoA domain, the N terminus of V 2 /IL␤ 2 .PhoA is most probably translocated Sec-independently. It was postulated that this is the general translocation mechanism for all N tail proteins without a signal peptide (26). In such Sec-independent translocation, charged residues are assumed to have a stronger topogenic potential, thus explaining the stronger influence of negatively charged residues in this case.
Whereas it seems clear that the electrochemical membrane potential can promote the translocation of negatively charged residues (7,8,11,12), they were thought to be topogenically active only under certain conditions. Among these were the presence of high numbers of negatively charged residues (9), a decreased hydrophobicity of the corresponding transmembrane domain (11), and the position-specific attenuation of positively charged residues lying in their conformational vicinity (10). For V 2 /IL␤ 2 .PhoA, a strong topogenic effect was seen even with a single glutamate residue. Taking the high hydrophobicity of the transmembrane domain of this receptor fragment into account, the first two conditions obviously do not apply here. A position-specific attenuation of positively charged residues may on first sight explain our results since a positively charged residue is present in the window region where the glutamate residues were introduced (Arg 38 ; see Figs. 1 and 5). However, attenuation can be ruled out, at least for this arginine residue, since its replacement by glutamine leads only to a relatively minor reduction of PhoA activity (72% of the wild type) (data not shown), demonstrating that Arg 38 is not essential to hold V 2 /IL␤ 2 .PhoA in the N in -C out orientation. Furthermore, the effect of glutamate residues was observed in an uninterrupted window region, and not at specific residue positions within a certain sequence, as was the case for negatively charged residues, which attenuate positively charged residues (10). Our results thus strongly suggest that negatively charged residues can play an active and direct role as topogenic determinants and support those of Kiefer et al. (12), who demonstrated that negatively charged residues help to translocate the N terminus of the Pf3 coat protein.
We have also shown that the topogenic potential of a single negatively charged residue depends on its distance from the transmembrane domain and have defined a window region of 6 residues. Since high resolution structural data for the TM1 boundaries of V 2 /IL␤ 2 .PhoA are not available, the length of this window region may vary by several residues. For a Pf3-Lep fusion protein, it was shown that when the hydrophobicity of the transmembrane domain of Lep was reduced, a negatively charged residue lying in the proximity of the transmembrane domain was necessary to translocate the N terminus, whereas a more distant residue was not (11). Although we observed the topogenic effects even with a highly hydrophobic transmembrane domain, these results suggest that a similar window region might exist for the N terminus of the Pf3 protein. The reason why single glutamate residues are effective only within a critical distance from the transmembrane domain remains obscure. The head groups of negatively charged phospholipids contribute to the retention of positively charged residues on the cytoplasmic side of the inner membrane (27). It is conceivable that these head groups interact only with residues in the proximity of the transmembrane domain. Repulsion of negatively charged residues on the cytoplasmic side may contribute to a mechanism that allows only those negatively charged residues lying within a critical distance from the transmembrane domain to influence the orientation of a membrane protein. On the other hand, the electrochemical membrane potential, which favors the translocation of negatively charged residues, and perhaps also their periplasmic retention, may also play a role. That the electrical field is of limited extension may explain the observed distance effects.