Analysis of the Role of the EnvZ Linker Region in Signal Transduction Using a Chimeric Tar/EnvZ Receptor Protein, Tez1*

Tez1 is a chimeric protein in which the periplasmic and transmembrane domains of Tar, a chemosensor, are fused to the cytoplasmic catalytic domain of EnvZ, an osmosensing histidine kinase, through the EnvZ linker. Unlike Taz1 (a similar hybrid with the Tar linker), Tez1 could not respond to Tar ligand, aspartate, whereas single Ala insertion at the transmembrane/linker junction, as seen in Tez1A1, restored the aspartate-regulatable phenotype. Analysis of the Ala insertion site requirement and the nature of the insertion residue on the phenotype of Tez1 indicated that a junction region between the transmembrane domain and the predicted helix I in the linker is critical to signal transduction. Random mutagenesis revealed that P185Q mutation in the Tez1 linker restored the aspartate-regulatable phenotype. Substitution mutations at Pro-185 further demonstrated that specific residues are required at this site for an aspartate response. None of the hybrid receptors constructed with different Tar/EnvZ fusion sites in the linker could respond to aspartate, suggesting that specific interactions between the two predicted helices in the linker are important for the linker function. In addition, a mutation (F220D) known to cause an OmpC c phenotype in EnvZ resulted in similar OmpC c phenotypes in both Tez1A1 and Tez1, indicating the importance of the predicted helix II in signal propagation. Together, (for EnvZ) in the of aspartate for Tez1A1/Tez3, conformation of the linker the relative positioning between domains and in A to a low kinase/phosphatase ratio of EnvZ, a low cellular concentration of OmpR-P. input (high osmolarity for EnvZ or aspartate for Tez1A1/Tez3), as piston/rotation/tilt or combinations of those motions are generated the transmembrane domains (16–21), the conformation of the junction region at the N-terminal end of the linker. the specific interactions between helices I and II in the linker to create conformational is propagated to helix AI of domain topological relationship between domains and to enzymatic (6), phosphatase ratio concentration of

EnvZ, a histidine kinase osmosensor, locates on the inner membrane of Escherichia coli. It is composed of a periplasmic domain, transmembrane domains, and a cytoplasmic domain (1,2). The cytoplasmic domain has been further dissected into three subdomains, the linker region, domain A, and domain B, in which the latter two form the catalytic core of EnvZ, harboring both kinase and phosphatase functions (3). NMR structures of both domain A and domain B have been solved (4,5). Biochemical analysis has revealed that the domain A of the EnvZ is responsible for the dimerization, phosphotransfer and phosphatase functions, and domain B binds ATP and catalytically assists the enzymatic function of domain A (3,6). The spatial arrangement between these two domains appears to be crucial for the modulation of EnvZ enzymatic activities (6). Previous studies suggest that EnvZ senses the extracellular osmolarity changes, transmits the signal through its transmembrane domain, and then modulates the kinase/phosphatase ratio of the cytoplasmic catalytic domain, which controls the cellular concentration of phosphorylated OmpR to mediate the reciprocal expression of the two major outer membrane porin proteins OmpF and OmpC (2,(7)(8)(9)(10)(11).
Although the exact ligand for EnvZ remains unknown, two kinds of chimeric receptors have been constructed in which the periplasmic and transmembrane domains of chemoreceptor Tar (sensor for aspartate) or Trg (sensor for ribose) are fused with the catalytic core of EnvZ (through the Tar or Trg linker, respectively). The resultant Taz (12) and Trz (13) are able to respond to aspartate and ribose in the medium, respectively, to activate ompC-lacZ in a concentration-dependent manner, suggesting that the chemoreceptors and EnvZ share a common signal transduction mechanism. Similar chimeric receptors have also been constructed using the periplasmic and transmembrane domain of Tar and the cytoplasmic domain of the human insulin receptor (14) or the periplasmic domain of the histidine kinase NarX and the cytoplasmic domain of Tar (15), further supporting a notion that there is a common mechanism widely used by signal transduction across the membrane in both prokaryotes and eukaryotes. Extensive studies on signal transduction through the transmembrane domain of chemotaxis receptors have been carried out, and various mechanisms have been proposed (16 -21). However, it remains unknown how the signal is propagated through the membrane to the cytoplasm. Particularly, it is of great interest how the linker region, a structural element connecting the transmembrane domain with the cytoplasmic signaling domain, propagates signal into the cytoplasm.
The linker region (also called HAMP domain) is widely found not only in histidine kinases and chemoreceptors but also in bacterial nucleotidyl cyclases and phosphatases. It is considered to function as a regulatory element in these proteins (22)(23)(24). Even though the linker regions share low primary sequence homology, they have a similar helix-turn-helix fold based on secondary structure prediction. A cysteine-scanning study has indicated that the Tar linker consists of a helix-turnhelix structure (25). However, no three-dimensional structures have yet been determined for any linker regions.
The EnvZ linker is also predicted to contain a helix-turnhelix structure (22,23,26) like the Tar linker. A number of mutations that block the osmosensing function of EnvZ have been mapped within the linker, suggesting its important role in signal transduction (26 -30). Only substitution of hydrophobic residue to charged residue, but not to hydrophobic residue, within the two predicted helices of the linker disturbed normal osmoregulation, suggesting that the amphipathic nature of the helices is important for signal propagation. It is important to note that the effects of the linker mutations on the function of the catalytic domain can be detected only when EnvZ is associated with the membrane (26). Once the cytoplasmic domain of EnvZ is detached from the membrane, the linker mutations no longer affect the catalytic function of EnvZ, indicating that the linker has to be fixed on the membrane to transmit the mutational effects to the catalytic domain.
To precisely elucidate the role of the EnvZ linker in signal propagation and the structural basis for its function, here we exchanged the Tar linker in the Taz construct with the EnvZ linker and examined how the linker exchange affects the aspartate-regulated EnvZ function. As shown in Fig. 1A, the sequence alignment between the EnvZ linker and the Tar linker together with the C-terminal regions of the transmembrane domain TM2 indicates that the EnvZ linker starts with one Arg residue, whereas the Tar linker starts with two Arg residues, resulting in an extra Arg residue at the transmembrane/linker junction for the Tar linker than the EnvZ linker. In the newly constructed Tar-EnvZ hybrid protein termed Tez1, the EnvZ linker starting with Arg-180 (the residue number is based on the EnvZ sequence) is fused to the Tar transmembrane domain TM2, ending at Ile-212 (the residue number is based on the Tar sequence). Tez1 could not induce ompC-lacZ in RU1012 cells even in the presence of aspartate. However, the insertion of a single Ala residue at the transmembrane/linker junction, not other regions in the linker, restored the aspartateregulatable signal transduction. In addition, through random mutagenesis we demonstrated that position 185 (the residue number is based on the EnvZ sequence), where a highly conserved Pro residue locates, is important for the linker function. To explore the possible structural basis for proper signal transmitting, we constructed several new Tar-EnvZ hybrids fused at different sites in the linker region. We also examined the effect of the F220D mutation, known to block EnvZ osmosensing, on Tez1 and its variant, Tez1A1. On the basis of the characterization of all these hybrid receptors and their mutants, we propose models as to how signal propagation through the linker region takes place.
Preparation of the Membrane Fraction-Membrane fractions containing Taz1, Tez1, or Tez1 variants were prepared as described previously (9). The protein amounts were determined by Bio-Rad protein assay. Western blot analysis was carried out using an equal amount of total membrane proteins (3 g).
Autophosphorylation and OmpR Kinase Assay-Membrane fractions containing receptor proteins (5 g) were incubated at room temperature in reaction buffer containing 50 mM Tris-HCl (pH 8.0), 50 mM KCl, 5 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 10 M ATP, and 0.5 Ci of [␥-32 P]ATP (3000 Ci⅐mmol Ϫ1 , 10 mCi⅐ml Ϫ1 ; PerkinElmer Life Sciences). Ten-l aliquots were removed from the reaction mixture at 2 and 15 min, and the reactions were stopped by adding 2.5 l of 5ϫ SDS loading buffer. In the remaining reaction mixture, OmpR was added to a final concentration of 2 M. Ten-l aliquots were removed at 5 and 10 min, and reactions were stopped as described above. All reaction mixtures were then analyzed by SDS-PAGE followed by autoradiography.
Screening for Intragenic Suppressor-Tez1 constructs harboring randomly mutagenized linker sequence were transformed into RU1012 cells. The screening for aspartate-regulatable constructs was carried out as described previously (33). Through this screening only one candidate (Tez2) was obtained. After confirming its phenotype by retransforming the mutated plasmid into RU1012 cells, the linker region of this plasmid was sequenced to identify the suppressor mutation.

Tez1 Cannot Respond to Aspartate, the Ligand for Tar-To
precisely examine the role of the linker region of EnvZ in signal transduction, we replaced the Tar linker (44 residues) in the Taz1 receptor with the EnvZ linker (43 residues) to construct a new hybrid protein termed Tez1, based on the assumption that the two linkers share a similar mechanism for signal propagation. In Tez1, the EnvZ linker was connected directly to the C-terminal end of the Tar transmembrane domain TM2 (Fig.  1A). As shown in Fig. 1B, Tez1 could not respond to aspartate and resulted in an OmpC-constitutive repression phenotype (OmpC Ϫ ) in RU1012 cells. Two possible reasons may account for this phenotype; (i) the linker replacement might compromise the stability of the receptor protein, since the lack of an aspartate receptor in the membrane is expected to result in the OmpC Ϫ phenotype even in the presence of aspartate, or (ii) the EnvZ linker is unable to transmit the input signal to the downstream catalytic domain, and thus, cells are unable to increase the OmpR-P concentration by increasing the kinase/ phosphatase ratio of the catalytic domain.
To test these possibilities, Western blot analysis of the membrane fractions from RU1012 cells harboring either Tez1 or Taz1 construct was carried out. As shown in Fig. 1C, Tez1 receptor could be expressed and localized in the membrane to a similar level as Taz1 receptor. Using the same membrane fractions containing Tez1 or Taz1 to check the autokinase and OmpR kinase functions of the two receptors, we found that Tez1, like Taz1, retained the enzymatic activities (Fig. 1D). The catalytic domain of EnvZ in the Tez1 receptor could autophosphorylate itself (lanes 1 and 2) and subsequently transfer the phosphoryl group to the OmpR (lanes 3 and 4) as that in Taz1 (lanes 5 and 6 for autophosphorylation; lanes 7 and 8 for phosphotransfer). These results suggest that the catalytic domain still retains its enzymatic functions but is unable to properly regulate the kinase/phosphatase ratio. Because the EnvZ linker in the intact EnvZ can propagate the osmolarity signal but not in Tez1, we speculate that a proper connection between the transmembrane domain and the linker region is essential to normal signal transduction.
Adjustment of the Transmembrane/Linker Junction in Tez1-The Tar transmembrane domain TM2 is presumed to consist of an ␣-helix structure (for review, see Ref. 17). The addition or deletion of an amino acid residue(s) at the transmembrane/linker junction may change the relative orientation of the EnvZ cytoplasmic domain to the Tar transmembrane domain through altering the conformation of the linker region, which in turn recovers the aspartate-regulatable OmpC expression in RU1012 cells. Therefore, we constructed three new Tez1 hybrid receptors, Tez1A1, Tez1⌬I181, and Tez1AA. In Tez1A1, one Ala residue was inserted immediately downstream of the membrane interface residue Arg-180; in Tez1⌬I181, Ile-181 residue was deleted; and in Tez1AA, two Ala residues were inserted after the 180 residue ( Fig. 2A). Note that at here and later sections, all the residue numbers used are based on the EnvZ sequence. As shown in Fig. 2B, Tez1A1 in RU1012 cells exhibited aspartate-regulatable ␤-galactosidase induction, as the expression of ompC-lacZ was well correlated to the medium aspartate concentration. In contrast, Tez1⌬I181 and Tez1AA could not respond to aspartate and displayed an OmpC Ϫ and an OmpC c phenotype, respectively. All three receptors were expressed and localized in the mem- Specificity of Amino Acid Insertion after Arg-180 -To further examine how the Ala residue inserted in Tez1A1 plays a specific role in signal transduction, random amino acid insertion was carried out after Arg-180 to construct Tez1X. Of 15 different residues introduced at this position, only Asp and Asn in addition to Ala residue were able to resume the aspartateregulatable regulation of ompC-lacZ in RU1012 cells as shown in Fig. 3. Notably, ␤-galactosidase activities in the absence of aspartate were quite high in the case of Asp and Asn residue insertion in contrast to Tez1A1. Amino acid residues such as Leu, which are preferred in helix formation, were not able to recover the aspartate-regulatable phenotype, suggesting that the Ala residue insertion after Arg-180 is not simply extending a helical structure through the TM2 transmembrane domain.
Most interestingly, the Tez1R construct, in which the EnvZ linker starts with Arg-Arg as the Tar linker and also contains 44 amino acid residues, exhibited an OmpC c phenotype. The fact that the residues to be inserted at the transmembrane/ linker junction are very selective indicates possible conformational requirements at the junction region for proper signal propagation.
Effects of an Ala Residue Insertion at Different Sites in the Linker-Next we examined whether length adjustment at other regions in the linker could recover aspartate-regulatable phenotype as seen in Tez1A1. Ala residues were inserted at three different positions (Tez1A2, Tez1A3, and Tez1A4) (Fig.  4A). In Tez1A2, an Ala residue was inserted in the predicted helix I, 7 amino acid residues downstream of the Tez1A1 fusion site, whereas in Tez1A3 an Ala residue was inserted in the predicted helix II. Note that according to the secondary structure prediction, helix II continues further into N-terminal end of the EnvZ catalytic domain A (22,23,26). In Tez1A4, an Ala residue was inserted immediately upstream of Arg-180. ␤-Galactosidase assays using RU1012 cells harboring these Tez1A variants showed that Tez1A2 and Tez1A3 exhibited an OmpC c phenotype even in the absence of aspartate, whereas Tez1A4 maintained an aspartate-regulatable phenotype as Tez1A1 (Fig. 4B). This suggests that the length adjustment in the linker for recovering signaling has to be at the transmembrane/ linker junction. The phenotypes of Tez1A2 and Tez1A3 also suggest that the helical orientations of the predicted two helices, helix I and helix II, play important roles in the linker function.

Important Role of Pro-185 in Signal
Transduction-To check whether any amino acid substitution mutations within the linker are able to achieve the same effect as the Ala insertion in Tez1A1 for aspartate-regulatable phenotype, PCR random mutagenesis was carried out on the entire linker region as described under "Experimental Procedures." After subcloning the mutagenized linker fragments of EnvZ (Arg-180 to His-222) back into the Tez1 construct, we transformed the plasmids into the RU1012 cells and plated the transformed cells on lactose-MacConkey plates containing 50 g/ml ampicillin and 5 mM aspartate. Subsequently, red colonies were picked to further test if their ␤-galactosidase activities were regulatable by aspartate.
Of more than 1500 red colonies, only one such mutant was found, which was termed Tez2. We confirmed the phenotype of Tez2 by extracting the plasmid from the cells, retransforming it back to the RU1012 cells, and assaying the ␤-galactosidase activities in the absence and presence of 5 mM aspartate. The ␤-galactosidase activity of Tez2 in the absence of aspartate is a little higher than that of Tez1A1; however, in the presence of 5 mM aspartate, ␤-galactosidase induction of Tez2 was almost as high as that of Tez1A1 (Fig. 5A). When the Tez2 linker region was sequenced, it was found to contain two amino acid substitutions, R184Q and P185Q. The two mutations were separated by site-directed mutagensis, and their individual phenotypes were examined. As shown in Fig. 5A, P185Q mutation (Tez3) alone could suppress the Tez1 phenotype and recover an aspartate-regulatable phenotype. On the other hand, the R184Q mutation (Tez4) was unable to suppress the Tez1 phenotype. Interestingly, when an Ala residue was inserted after Arg-180 in Tez3, as seen in Tez1A1, the resultant Tez3A was still able to show an aspartate-regulatable phenotype (Fig. 5A), indicating that the P185Q mutation overrides or bypasses the length effect of the junction region or vice versa.
To further demonstrate the importance of position 185 in signal transduction, we carried out random mutagenesis at this location. As shown in Fig. 5B, three different phenotypes were observed depending on the identity of the amino acid substituted; that is, aspartate-regulatable (Glu and Ile), partially regulatable (Asn and Leu), or non-regulatable (Arg, Ser, Asp, and Ala). Western blot analysis demonstrated that all mutant proteins were expressed and localized in the membrane fractions at a similar level as Tez1 with the exception of Pro to Ser mutation, which resulted in about a 50% reduction in expression (data not shown). All the mutant proteins also retained enzymatic activities (data not shown).
These data demonstrate that position 185 plays a critical role in the EnvZ linker function. It should be noted that Pro-185 is one of the most conserved residues within the linker in all chemoreceptors and many histidine kinases in E. coli (23,24,26). That a Pro residue rather than other amino acid residues is chosen in native receptor proteins may be due to the special requirement of the linker conformation with respect to the transmembrane domains.
Specific Interactions between helix I and helix II in the Linker-Both Tar and EnvZ linkers could serve as a functional unit in signal transduction. Although they share little primary sequence homology, a similar helix (helix I)-turn-helix (helix II) structure fold is adapted by the two linkers, which may underline a similar structural basis for signal transduction. To examine whether the conservation of helical amphipathicity or specific residue interactions are required for the linker function, two different hybrid receptors were constructed in which helix I from the Tar linker and helix II from the EnvZ linker were connected in two different manners in the loop region as shown in Fig. 6A. Both constructs (Tez5 and Tez6) exhibited OmpC c phenotypes, which are similar to that of a previously characterized Tar/EnvZ hybrid receptor, Taz2-1 (9) (Fig. 6). We also constructed three other Tar/EnvZ hybrid receptors, Tez7, Tez8, and Tez9, in which the fusions were made within helix II (Fig. 6A). Unlike Tez5, Tez6, Taz2-1, and Taz1, these hybrid constructs exhibited OmpC Ϫ phenotypes even in the presence of 5 mM aspartate (Fig. 6B). Therefore, both helix I and helix II in the linker have to be derived from the same protein for the normal linker function. Specific interactions between the two helices in the linker seem to be critical to signal transduction, although further investigation is needed to pinpoint the residues that are involved in the interactions.

Effects of F220D Mutation on the Phenotypes of Tez1A1 and Tez1
Receptors-Based on the LEARNCOIL analysis, part of helix II of the EnvZ linker (residues 216 -222) may continue into the downstream helical region of the EnvZ domain A to form a coiled-coil structure in a dimer configuration. A similar prediction has been made for other histidine kinases, suggesting that a coiled-coil structure may be a common feature for most of the histidine kinases (34). In helix II of EnvZ, it has been shown that substitution of a hydrophobic to a charged amino acid residue at position 220 (F220D) blocked EnvZ osmolarity sensing, resulting in an OmpC c phenotype (26). One may speculate that this mutation fixes the helical orientation of the possible coiled-coil structure; as a result, the spatial arrangement between the downstream catalytic domains A and B is no more regulatable by external signals. The kinase/phosphatase ratio of EnvZ becomes fixed and in turn results in a non-regulatable phenotype. Indeed, after introducing the same mutation into the Tez1 and Tez1A1 constructs, both Tez1A1 (regulatable) and Tez1 (non-regulatable; OmpC Ϫ ) exhibited OmpC c phenotypes (Fig. 7), indicating that the F220D mutation overrules host phenotypes to convert the downstream catalytic domains in a kinase ϩ /phosphatase Ϫ conformation.

FIG. 5. Important role of Pro-185 in signal transduction.
A, intragenic suppressors of Tez1. PCR random mutagenesis in the entire EnvZ linker region followed by screening for aspartate-responsible mutants were carried out as described under "Experimental Procedures." A suppressor mutant thus isolated (Tez2) harbored suppressor mutations P185Q and R184Q. These two mutations were separated by site-directed mutagenesis, and the two Tez mutants were named Tez3 (Tez1 harboring the P185Q mutation) and Tez4 (Tez1 harboring the R184Q mutation). An Ala residue was inserted after Arg-180 within Tez3, resulting in Tez3A. ␤-Galactosidase activities of Tez2, Tez3, Tez4, and Tez3A as well as Tez1 and Tez1A1 were measured with (black bar) or without (hatched bar) 5 mM aspartate in the growth medium. B, the ␤-galactosidase activities of E. coli RU1012 cells carrying various Tez1P185X mutants. Random mutagenesis at Pro-185 was carried out using Tez1 as template. ␤-Galactosidase activities of various Tez1P185X receptors were measured with (black bar) or without (hatched bar) 5 mM aspartate in the growth medium. The data plotted were the average values from three independent experiments.

DISCUSSION
In the present study, the phenotypes of Tez1 and its variants in comparison to that of Taz1 suggest that although the EnvZ linker and the Tar linker share similar secondary structures, they need to adapt proper phase configurations with respect to the upstream transmembrane domains and the downstream catalytic domains for signal transduction. The transmembrane/ linker junction is crucial for the linker function, as present results showed that adjustment at the junction resulted in all possible signaling modes: the constitutive "off " phenotype, the ligand-responsive "off " to "on" regulatory phenotype, and the constitutive "on" phenotype. The phenotypes of Tez1X further indicated that a specific residue is needed at the junction for proper signal transduction in addition to the possible length requirement for the linker function. Because both the EnvZ linker and the Tar linker are able to propagate external signals to regulate the function of the downstream EnvZ catalytic domain, it seems evident that EnvZ and Tar share a similar mechanism for signal transduction.
Insertion of an Ala residue at different sites in the EnvZ linker led to different phenotypes for Tez1, indicating that the length requirement is not the only determining factor for the linker function. The aspartate responsiveness of Tez1A1 and Tez1A4 further confirms the importance of the transmembrane/linker junction. In addition, the phenotypes of Tez1A1 and Tez1A2, in which an Ala residue was inserted seven residues apart, suggest that unlike what has been proposed for the Tar linker (25), a junction region (Arg-180 to Leu-186) may exist at the N-terminal end of the EnvZ linker that does not simply adapt a helical structure connecting the transmembrane domain TM2 with the predicted helix I of the EnvZ linker to form a single long helical structure. Within this junction region, the residue at position 185 seems to be crucial for the linker function, because substitution mutations at Pro-185 lead to all the three possible phenotypes for Tez1. Note that it has been shown that substitution mutations at Arg-180 (R180C and R180W) in EnvZ resulted in OmpF ϩ OmpC Ϫ phenotypes (11,30), and the P185L mutation in EnvZ led to an OmpC c phenotype (11), again confirming the notion that the junction  (35). Helix AI also forms a four-helix bundle structure with helix II of domain A (helix AII; shown by rods) in a dimer (4). The C-terminal end of domain A is further connected to domain B (shown as kidney-shaped structures). B, signal propagation in the cytoplasmic domain of EnvZ upon signal input. Under the low osmolarity (for EnvZ) or in the absence of aspartate for Tez1A1/Tez3, the conformation of the linker region keeps the relative positioning between domains A and B as seen in A to maintain a low kinase/phosphatase ratio of EnvZ, leading to a low cellular concentration of OmpR-P. Upon signal input (high osmolarity for EnvZ or aspartate for Tez1A1/Tez3), motions such as piston/rotation/tilt or combinations of those motions are generated across the transmembrane domains (16 -21), which alter the conformation of the junction region at the N-terminal end of the linker. This then changes the specific interactions between helices I and II in the linker to create conformational signal, which is further propagated to helix AI of domain A. In turn, the topological relationship between domains A and B is altered to reset the enzymatic functions of EnvZ (6), allowing a high kinase/ phosphatase ratio to result in a high cellular concentration of OmpR-P. region is crucial for signal transduction. The aspartate-regulatable phenotype of Tez3A, which contains an Ala residue insertion at the transmembrane/linker junction together with the P185Q mutation in Tez1 (the residue number is based on the EnvZ sequence) further implies that the junction region functions as a single structural unit for modulating the conformation of the EnvZ linker. Specific amino acid requirements, as seen in Tez1X and Tez1P185X mutants, together with the fact that the linker has to be fixed on the membrane for its proper function indicates that the interaction between the membrane and the junction region is crucial for proper signal propagation. Alternatively, because both Tar and EnvZ have an unusually long transmembrane TM1 region, possible interactions between the junction region and the portion of TM1 facing the cytoplasmic space under different signaling conditions may also play an important role in modulating the linker conformation.
Consistent with the previous observation (26), the helical orientations of both helix I and helix II in the EnvZ linker may be important for the linker function, since the Ala insertion in either helix I or helix II locked signal transduction at the "on" signaling mode. The phenotypes of different Tar-EnvZ hybrid receptors in the present paper (Tez5, -6, -7, -8, and -9) as well as those of Taz1 and Taz2-1 in the previous works (9,12) suggest that both the Tar linker and the EnvZ linker need to be intact for fulfilling their function, and specific interactions between helix I and helix II in the linker is important for signal transduction. Particularly, helix II may serve as an effector element within the linker for further signal propagation, as it may form a long coiled-coil structure together with helix I of domain A of EnvZ based on LEARNCOIL prediction. The F220D mutation in helix II of either the EnvZ, Tez1, or Tez1A1 linker causes an OmpC c phenotype in all cases, implying that mutations at helix II of the linker alter the helical orientation of the predicted coiled-coil structure so that the enzymatic functions of the downstream catalytic domain of EnvZ can no longer be modulated by external signals. It has been shown that signal transduction in several receptor proteins is regulated by altering the coiled-coil orientation (20,35).
On the basis of the present work and previous studies one may speculate how the signal input from the transmembrane domains is propagated through the linker region to modulate the enzymatic functions of downstream catalytic domains A and B of EnvZ. As shown in Fig. 8A, in the cytoplasmic portion of EnvZ, the junction region (including the conserved Pro residue at position 185) at the N-terminal end of the EnvZ linker and the specific interactions between two predicted helices (I and II) within the linker play important roles in signal transduction. Also, as mentioned above, helix II of the linker is predicted to form a long coiled-coil structure together with helix I of domain A of EnvZ, serving as an effector element for further signal propagation. Downstream of the linker region, domains A and B retain the enzymatic activities of EnvZ, and the relative positioning of these two domains is crucial for EnvZ function modulation (6,33). We propose that under the low osmolarity conditions (for EnvZ) or in the absence of aspartate in the medium (for Tez1A1 or Tez3), the conformation of the linker region determines a specific relative positioning between domains A and B so that the kinase/phosphatase ratio of EnvZ is maintained at lower levels, resulting in lower cellular concentrations of OmpR-P to activate the ompF gene transcription (Fig. 8A). On the other hand, under the high osmolarity condition (for EnvZ) or in the presence of aspartate in the medium (for Tez1A1 or Tez3), the external signals are converted into molecular displacements between the two subunits in a dimer in forms of piston motion, rotation, tilt, or the combination of those motions across the transmembrane domains (16 -21).
The conformation of the junction region is, thus, altered to modulate the specific interactions between helices I and II within the linker. Helix II of the linker further propagates the conformational signals to helix I of domain A through the long coiled-coil structure. As a result, the relative positioning between domains A and B is changed to reset the kinase/phosphatase ratio of EnvZ to a higher value, which results in higher cellular concentrations of OmpR-P to activate the ompC gene transcription and to repress the ompF gene transcription (Fig. 8B).
In the model presented in Fig. 8, the two helices in the linker together with their counterparts within a receptor dimer are presumed to form a parallel four-helix bundle similar to those observed in the basic helix-loop-helix structure of transcription factors such as Max and Myc (36 -39). The fact that the linker function requires its attachment to the membrane indicates that the cytoplasmic membrane may be essential for maintaining a stable four-helix bundle structure in the linker. The junction region in respect to the transmembrane domains may also be important for the linker structure and the relative helical alignment within the linker four-helix bundle. However, it is still possible that helix I of the linker lies parallel to the plane of the membrane surface, which dissociates from the membrane to transiently interact with helix II in the linker upon signal input as proposed by Williams and Stewart (24). In that case, the relative flexibility of the junction region may also allow the N-terminal regions of helix I to interact with each other in an antiparallel fashion. Further structural studies are needed to distinguish those possibilities.