The identification of a minimal dimerization motif QXXS that enables homo- and hetero-association of transmembrane helices in vivo.

Assembly of transmembrane (TM) domains is a critical step in the function of membrane proteins, and therefore, determining the amino acid motifs that mediate this process is important. Studies along this line have shown that the GXXXG motif is involved in TM assembly. In this study we characterized the minimal dimerization motif in the bacterial Tar-1 homodimer TM domain, which does not contain a GXXXG sequence. We found that a short polar motif QXXS is sufficient to induce stable TM-TM interactions. Statistical analysis revealed that this motif appears to be significantly over-represented in a bacterial TM data base compared with its theoretical expectancy, suggesting a general role for this motif in TM assembly. A truncated short TM peptide (9 residues) that contains the QXXS motif interacted slightly with the wild-type Tar-1. However, the same short TM peptide regained wild-type-like activity when conjugated to an octanoyl aliphatic moiety. Biophysical studies indicated that this modification compensated for the missing hydrophobicity, stabilized alpha-helical structure, and enabled insertion of the peptide into the membrane core. These findings serve as direct evidence that even a short peptide containing a minimal recognition motif is sufficient to inhibit the proper assembly of TM domains. Interestingly, electron microscopy revealed that above the critical micellar concentration, the TM lipopeptide forms a network of nanofibers, which can serve for the slow release of the active lipopeptide.

The function of many proteins can be regulated by alteration of their oligomerization state. For example, information that the cells receive from the outside is thought to be communicated to cells via changes in the oligomerization of membrane receptor proteins. This information is crucial for important cellular processes such as homeostasis and signal transduction (1)(2)(3)(4)(5)(6). Receptor oligomerization is mainly mediated by the extracellular or intracellular domains. However, considerable data have been accumulated concerning the causal involvement of the transmembrane (TM) 1 domains in this process as well (7)(8)(9)(10)(11). In contrast to the soluble regions of membrane proteins, our knowledge of the factors that control proteinprotein interaction and recognition of the membrane-embedded domains is still limited.
The only TM domain dimerization motif identified to date is the GXXXG sequence (11)(12)(13)(14). However, there are many examples of TM domains that assemble even though they do not contain the GXXXG motif. One such example is the N-terminal TM domain of the Escherichia coli aspartate receptor (Tar). This receptor is one of the main chemotaxis receptors found in bacteria, and it forms a homodimer complex in which each subunit is composed of two TM helices (Tar-1 and Tar-2) separated by a substantial periplasmic domain. Biochemical studies demonstrated that the N-terminal TM domain of the receptor (Tar-1) is able to dimerize and therefore contributes to the dimerization of the protein (15)(16)(17), although dimerization is mainly driven by its periplasmic or cytoplasmic domains (15, 18 -20).
Previously, we studied the dimerization of the Tar-1 TM domain by grafting it into the ToxR TM system, which can assess the association level of a particular TM domain within the E. coli natural membrane (21). Using this system, we identified a polar residue motif, QXXS, which is crucial for the interaction and the recognition of the Tar-1 TM domain in vivo (17). It was suggested that this motif stabilizes the dimerization of Tar-1 through the formation of inter-helical hydrogen bonds. In support of this, a dominant-negative effect was observed in the presence of a synthetic peptide corresponding to the Tar-1 wild-type TM domain, but not in the presence of a mutant peptide in which the two polar residues were substituted for two non-polar residues (22).
The significance of polar residues in the association of TM domains has been previously studied both in vitro, by examining TM domain synthetic peptides with polar residues in their sequence, and in vivo, by analysis of de novo designed TM helices (23)(24)(25)(26). These studies revealed that amino acids containing two polar side chain atoms (such as asparagine and glutamine) have a greater tendency to drive TM association than residues containing only one polar side chain atom (threonine or serine) (23)(24)(25)(26). However, the presence of polar residues within the TM domain is not always sufficient for TM assembly (27). Moreover, there are well documented examples of non-covalent associations of TM segments that are mediated merely by van der Waals packing interactions (11)(12)(13)(14).
In this study we investigated whether the polar residue motif (QXXS) is sufficient by itself to stabilize the TM-TM assembly of the native Tar-1 TM domain, or whether the specific amino acid surrounding it has an additive or a synergistic effect on assembly. For this purpose we examined: (i) whether the short QXXS motif can induce dimerization of non-dimerizing TM domains such as leucine and alanine backbones, and (ii) the dominant-negative effect of four truncated TM peptides on the dimerization of wild-type Tar-1. Furthermore, to determine whether the polar residue dimerization motif (QXXS) is a general motif that appears in additional TM domains, we statistically analyzed the frequency of occurrence of the motif in a large set of bacterial TM sequences. Our data reveal a minimal dimerization motif sufficient for the assembly of the Tar-1. This motif is highly over-represented in a bacterial TM domain data base constructed from Swiss-Prot.

MATERIALS AND METHODS
Construction of the ToxR Chimeras-A NheI-BamHI TM-DNA cassette encoding 16 residues of the Tar-1 WT TM domain ( 13 MVLGV-FALLQLISGSL 28 ) was inserted between the ToxR transcription activator and the E. coli maltose-binding protein (MalE) within the ToxR-MalE plasmid. The 9TMϩLeu construct contained nine residues of the WT TM domain sequence (positions 20 -28) and a sequence of eight leucines. The QXXS motifϩbackbone and QXXS motifϩAla constructs are composed of the dimerization motif ( 22 QXXS 25 ) within a stretch of leucine and alanine, respectively (Table II). The sequence of all the constructs was confirmed by DNA sequencing.
Peptide Synthesis and Purification-Peptides were synthesized by the Fmoc solid-phase method on a Rink amide MBHA resin. The lipophilic acid was attached to the N terminus of a resin-bound peptide using standard Fmoc chemistry. Briefly, after removal of the Fmoc from the N terminus of the peptide with a solution of 20% piperidine in dimethylformamide, the fatty acid (7 eq, 1 M in dimethylformamide) was coupled to the resin under similar conditions used for the coupling of an amino acid. The peptides were cleaved from the resin by trifluoroacetic acid and were purified by reverse phase-high performance liquid chromatography on a C4 reverse phase Bio-Rad semi-preparative column (250 ϫ 10 mm, 300 Å pore size, 5-m particle size). The purified peptides were shown to be homogeneous (Ͼ95%) by analytical high performance liquid chromatography. The composition of the peptides were confirmed by electrospray mass spectrometry. Lysine residues were added to the C termini of the peptides (and in the case of Tar-1 16TM peptide also to the N termini) to confer water solubility to the hydrophobic TM domains (28,29). It was previously shown that hydrophobic peptides conjugated to lysine tags were correctly oligomerized and inserted into the membrane (28 -30).
In Vivo Detection of Homo-and Heterodimerization of TM Domains within the Membrane-The ToxR transcription activator can be used successfully to assess weak protein-protein interactions within the E. coli membrane. A Tar-1 TM domain encoding the DNA cassette was grafted between the ToxR transcription activator and the maltosebinding protein in the ToxR-MalE plasmid. The plasmid was then transformed into E. coli FHK12 cells, which contain ␤-galactosidase, under the control of a ctx promoter. Dimerization of the TM domains, in this system, results in association and activation of the ToxR transcription activator, which then becomes active and is able to bind the ctx promoter (21). Quantification of the amount of homodimerization was done by measuring the activity of the ␤-galactosidase reporter gene and by normalizing it to the cell content (A 590 ) (miller units). The baseline activity of a negative control ToxRЈA 16 , which remains a monomer, was subtracted from all the results (21). The transformed cells were grown in the presence of chloramphenicol for 18 h at 37°C. ␤-Galactosidase activities were quantified in crude cell lysates after adding o-nitrophenylgalactosidase and monitoring the reaction at 405 nm for 20 min, at intervals of 30 s at 28°C with a Molecular Devices kinetic reader (21,31). Specific ␤-galactosidase activities were computed from the V max of the reaction.
Hetero-association was detected using ToxR-Tar-1-expressing bacteria grown in the presence of exogenous peptides. Inhibition was calculated as follows, where I represents the inhibitory ability of the peptide, A peptide is the activity of ToxR-Tar-1 in the presence of peptide, A max is the maximal activity of ToxR-Tar-1 without the peptide, and A baseline is the baseline activity of the monomer A 16 plasmid (21). ToxR-TM-MalE Chimera Protein Expression Levels-We performed Western blot analysis to determine whether change in the sequence of the TM domain or the presence of the peptides affected the expression levels of the chimera protein. Specifically, aliquots of 10 l of FHK12 cells, each with a different plasmid or in the presence of a different peptide, were mixed with a sample buffer, boiled for 5 min, subjected to 12% SDS-PAGE, and then transferred to nitrocellulose. The primary antibody used was anti-maltose-binding protein. The detection was done with a "Phototope-HRP Western blot Detection System" from Cell Signaling Technology.
Maltose Complementation Assay-Membrane insertion and correct orientation were examined as previously described (32). Briefly, PD28 cells, transformed with the different plasmids, were cultured overnight. The cells were then washed twice with PBS and used to inoculate M9 minimal medium including 0.4% maltose at a 200-fold dilution. The growth of the cells was measured at different times by cell density at 650 nm.
Binding Analysis by Using BIAcore Biosensor-Surface plasmon resonance (SPR) experiments were carried out with a BIAcore 3000 analytical system (Biacore, Uppsala, Sweden) using an L1 sensor chip (Biacore). The L1 sensor chip is composed of carboxylmethylated dextran matrix modified with lipophilic substances covalently linked to a gold surface. This chip is designed to capture lipid bilayers vesicles. The running buffer used for all the experiments was PBS without Ca 2ϩ and Mg 2ϩ (pH 6.8). The washing solution was 40 mM N-octyl-␤-D-glucopyranoside. All solutions were freshly prepared, degassed, and filtered through 0.22-m pores. All the experiments were done at a temperature of 25°C. After cleaning the system according to the manufacturer's instructions, the BIAcore 3000 instrument was left running overnight using Milli-Q water as eluent to thoroughly wash all liquid handling parts of the instrument. The L1 chip was then installed and the surface was cleaned by an injection of the non-ionic detergent, 40 mM N-octyl-␤-D-glucopyranoside (25 l), at a flow rate of 5 l/min. PE/PG (7:3, w/w) vesicles (80 l, 0.5 mM) were then applied to the chip surface at a low flow rate (2 l/min). To remove any multilamellar structures from the lipid surface, NaOH (50 l, 10 mM) was injected at a flow rate of 50 l/min, which resulted in a stable baseline corresponding to the lipid monolayer linked to the chip surface. This bilayer, linked to the chip surface, was then used as a model membrane surface to study peptide-membrane binding. In a typical experiment, peptide solutions (25-l PBS solution of 0.1-10 M peptide) were injected on the lipid surface at a flow rate of 5 l/min. PBS alone then replaced the peptide solution for 1200 s to allow for peptide dissociation.
ATR-FTIR Measurements-Spectra were obtained with a Bruker equinox 55 FTIR spectrometer equipped with a deuterated triglyceride sulfate detector that was coupled to an ATR device. For each spectrum, 150 scans were collected, with a resolution of 4 cm Ϫ1 . Samples were prepared as previously described (33). Briefly, a mixture of PE/PG (1 mg) alone or with peptide was deposited on a ZnSe horizontal ATR prism (80 ϫ 7 mm). The aperture angle of 45°yielded 25 internal reflections. Prior to preparing the samples, the trifluoroacetate (CF3COO Ϫ ) counterions, which strongly associate with the peptide, were replaced with chloride ions by washing in 0.1 M HCl and lyophilization. This eliminated the strong C ϭ O stretching absorption band near 1673 cm Ϫ1 (34). Lipid/peptide mixtures were prepared by dissolving the peptide in trifluoroethanol and the lipid in a 1:2 MeOH: CH 2 Cl 2 mixture and drying them under vacuum for 15 min. Lipid/ peptide mixtures or lipids were spread with a Teflon bar on the ZnSe prism. Polarized spectra were recorded, and the respective pure phospholipids in each polarization were subtracted to yield the difference spectra. The background for each spectrum was a clean ZnSe prism. Deuterium exchange was carried out as previously described in detail (35).
ATR-FTIR Data Analysis-Prior to curve fitting, a straight base line passing through the ordinates at 1700 and 1600 cm Ϫ1 was subtracted. To resolve overlapping bands, the spectra were processed using PEAK-FIT TM (Jandel Scientific, San Rafael, CA) software. Second-derivative spectra accompanied by 13-data point Savitsky-Golay smoothing were calculated to identify the positions of the component bands in the spectra. These wavenumbers were used as initial parameters for curve fitting with Gaussian component peaks. Positions, band widths, and amplitudes of the peaks were varied until: (i) the resulting bands shifted by no more than 2 cm Ϫ1 from the initial parameters, (ii) all the peaks had reasonable half-widths (Ͻ25 cm Ϫ1 ), and (iii) good agreement between the calculated sum of all components and the experimental spectra was achieved (r 2 Ͼ 0.999). The relative contents of different secondary structure elements were estimated by dividing the areas of individual peaks, which were assigned to a particular secondary structure, by the whole area of the resulting amide I band. The results of three independent experiments were averaged.
Circular Dichroism (CD) Spectroscopy-The CD spectra of the peptides were measured in an Aviv 202 spectropolarimeter. The spectra were scanned with a thermostated quartz optical cell with a path length of 1 mm. Each spectrum was recorded at 1-nm intervals with an average time of 10 s, at a wavelength range of 260 -190 nm. The peptides were scanned at a 100 M concentration in 1% lipophosphatidylcholine micelles. Fractional helicities (36,37)  , corresponding to 0 and 100% helix content at 222 nm, are estimated to be Ϫ2,000 and Ϫ32,000 deg cm 2 /dmol, respectively (36).
ANS Fluorescence Measurements-8-Anilinonaphthalene-1-sulfonate (ANS) is an environmentally sensitive fluorescence probe, which we used to detect the critical micelle concentration (CMC) of the lipopeptide 9WTϩoct. ANS is known to have a very low fluorescence yield in an aqueous environment. The fluorescence yield increases significantly upon transfer to a hydrophobic environment (38). The peptides at different concentrations were mixed with 10 M ANS and the fluorescent emission was measured at room temperature using the SLM-Aminco Series 2 Spectrofluorimeter with excitation set at 350 nm and emission wavelength of 530 nm (16 nm slit).
Visualization of Peptides by Electron Microscopy-A drop containing 100 M peptide in double distilled water was deposited onto a carboncoated grid and negatively stained with 2% phosphotungstic acid (pH 6.8). The grids were examined using a JEOL JEM 100B electron microscope (Japan Electron Optics Laboratory Co., Tokyo, Japan).
Statistical Analysis-The analysis of the frequencies of occurrence of the QXXS, QXXXS, and GXXXG motifs were performed on a bacterial TM domain data base. The source of the TM sequences was the annotated Swiss-Prot data base updated to February 2005. The data base contains 41,916 bacterial TM domains with length ranging between 15 and 30 amino acids. The occurrences of the QXXS, QXXXS, and GXXXG motifs in the data base were counted. The averaged expected frequency of occurrences was calculated by counting the number of motif occurrences in 100 randomized data bases. Randomization of the data base was achieved by combining TM domain sequences of a particular length into one long string of characters, which was then shuffled randomly and re-cut into the original length.

Homodimerization of Different TM Domains That
Contain the QXXS Motif-The ToxR assembly system can detect TM-TM interactions within the inner membrane of E. coli. The dimeric state of the chimera protein is indicated by the activity (V max ) of the ␤-galactosidase reporter gene. A ToxR-MalE chimera protein with a TM domain that contains the C-terminal half of the Tar-1 and a stretch of eight leucines (9TMϩLeu) was examined for its self-association ability to determine whether the C-terminal segment, which contains the QXXS motif, is sufficient for TM-TM interaction ( Table I).
The 9TMϩLeu construct showed dimerization ability similar to that of the construct that contains the wild-type TM domain (Fig. 1A). This result implies that the full dimerization motif is localized on the C-terminal half of the Tar-1. The role of the N-terminal segment is probably to confer hydrophobicity and sufficient length for the TM domain to traverse the membrane, and it can therefore be replaced by a leucine stretch.
To determine whether the QXXS sequence is the minimal motif sufficient for TM-TM dimerization, we examined the dimerization activity of two additional constructs. These constructs contain the QXXS motif either in the context of a   Table I. B, comparison of the expression levels of the chimera protein ToxR-TM-MalE (65 kDa). Samples of FHK12 cells containing different sequences of Tar-1 within the ToxR-MalE chimera protein were lysed in sodium dodecyl sulfate sample buffer, separated on 12% SDS-PAGE, and immunoblotted using anti-maltose-binding protein antibody (New England Biolabs). The chimera protein mutants showed expression levels similar to the WT TM domain or higher. C, correct integration of the ToxR-TM-MalE chimera proteins was examined by their ability to functionally complement the MalE deficiency of PD28 cells. PD28 cells were transformed with Tar-1 WT (‚), 9TMϩLeu (f), QXXS motif ϩ backbone (छ), only backbone (OE), QXXS motifϩAla (F), and ⌬TM (Ⅺ) plasmids, and were grown in minimal medium containing maltose. QXXS motif ϩbackbone, only backbone and 9TMϩLeu constructs showed growth curves similar to Tar-1 WT, indicating proper membrane integration. The QXXS motifϩAla construct and the negative control with the deleted TM domain (⌬TM) showed no growth. stretch of leucines and alanines or only an Ala backbone (the exact sequences are indicated in Table I). These backbones were chosen because both have a low dimerization propensity and inserted properly within the membrane (ToxR-A 16 construct severs as a marker for baseline activity for all the experiments). The QXXS motif ϩ backbone showed a dimerization level similar to the Tar-1 WT TM domain, whereas the backbone alone showed very low dimerization activity. However, the QXXS motifϩAla showed no dimerization ability at all. To test whether the remarkable difference between the dimerization activities of the two constructs was the result of low expression levels of the chimera proteins, or alternatively, from a failure of the Ala construct to properly insert into the membrane, we performed Western blotting and maltose complementation assays (Fig. 1, B and C, respectively). The expression levels of the chimera proteins were similar or higher than the Tar-1 WT. Thus, we can exclude the possibility that the low dimerization ability of the QXXS motifϩAla is the result of a lower chimera-protein expression level. Note that all activities were normalized to the protein expression level, using the Western blotting results.
Correct integration of the ToxR-TM-MalE chimera proteins into the inner membrane of E. coli was determined by examining the ability of the mutants to functionally complement a MalE-deficient E. coli strain (PD28) (32). Because PD28 cells are unable to grow on minimal medium with maltose as the only carbon source, only cells that express the chimera protein in the right orientation (MalE pointed toward the periplasm) will be able to utilize maltose and thus will allow cell growth. The QXXS motif ϩ backbone, backbone alone, and 9TMϩLeu constructs showed cell growth rates similar to Tar-1 WT, indicating proper membrane integration (Fig. 1C). However, the QXXS motifϩAla construct showed no growth up to 48 h. A construct with a deleted TM domain (⌬TM) served as a negative control and should reside in the cytoplasm. The control was unable to complement the MalE deficiency, as expected (Fig.  1C). These results indicate that the QXXS motifϩAla chimera protein is unable to insert into the membrane and therefore it has no dimerization activity.
The Dominant Negative Effect of Tar-1 Peptides on TM-TM Assembly-Heterodimerization was examined by monitoring the dominant-negative effect of Tar-1 peptides on the assembly of the Tar-1 WT TM domain using the ToxR-MalE system ( Fig.  2A) (22, 39 -43). To detect the minimal dimerization motif that is sufficient for heterodimerization, we examined the ability of four different peptides corresponding to different lengths of the Tar-1 TM domain to inhibit the proper Tar-1 dimerization. Only peptides that contain the full dimerization motif will be able to inhibit the dimerization activity to a similar extent as the full-length TM peptide. The sequences and designations of the peptides are shown in Table II. The exogenous Tar-1truncated peptides (20 M) showed a length-dependent inhibition ability when introduced to the ToxR-Tar-1-MalE chimera protein (Fig. 2B). The inhibition activity of the Tar-1 3WT peptide represents a nonspecific interaction, because this peptide does not contain the QXXS motif essential for Tar-1 dimerization.
The length-dependent inhibition of the peptides can be explained by the fact that truncation of the peptides either eliminated critical amino acids needed for dimerization or reduced the total hydrophobicity of the peptides below the threshold needed for their proper insertion into the membrane. To distinguish between these two possibilities, we compared the inhibition ability of the 9WT peptide, which displayed moderate inhibition activity, and its octylated analog (9WT ϩ octanoic acid). Conjugation of a fatty acid moiety should compensate for the hydrophobicity lost as a result of TM truncation. Tar-1 9WTϩoct showed an inhibition profile similar to the Tar-1 16WT peptide (Fig. 3A), indicating that both peptides interfered with the proper dimerization of the chimera protein to the same extent. These results demonstrate that the reduced inhibition activity of the Tar-1 9WTtruncated peptide is because of insufficient hydrophobicity and/or an inappropriate secondary structure, and not because of the loss of critical amino acids.
To rule out the possibility that the peptides directly reduced The C-terminal is amidated. b Lysine residues were added to confer water solubility (28,29).
the expression of the chimera protein and not the TM-TM interaction, we analyzed the expression levels of the chimera proteins in the presence of 20 M of each peptide. Western blotting confirmed that the peptides had no effect on the expression levels of the chimera proteins, because they remained constant in all samples (Fig. 3B). The heterodimerization result observed for the Tar-1 9WTϩoct peptide complements the homodimerization results observed for the 9TMϩLeu and the QXXS motifϩLeu constructs. Together, these results suggest that the QXXS motif is the minimal motif sufficient for driving the assembly of the Tar-1 TM domain (Fig. 1).
Peptide Binding to PE/PG Bilayer, as Determined by Surface Plasmon Resonance-Membrane binding and integration of the Tar-1 9WT and 9WTϩoct peptides were determined using SPR on a PE/PG (7:3 w/w) bilayer, a phospholipid composition typical of E. coli (44), using the L1 chip. The binding sensograms of the peptides revealed different binding responses (Fig. 4). The 9WT peptide showed reversible binding and reached a steady-state, whereas its octylated analog displayed slower kinetics and did not reach a steady-state. These results imply that conjugation of octanoic acid to the TM peptide affects its PE/PG binding properties. Apparently, the 9WT peptide binds to the surface of the phospholipid membranes probably through electrostatic interaction; however, it is not hydrophobic enough to insert into the membrane core. The Tar-1 9WTϩoct peptide, on the other hand, which is more hydrophobic, shows slower binding kinetics and incomplete dissociation. Its slow and incomplete dissociation profile suggests insertion into the hydrophobic core of the membrane.
Secondary Structure of the Peptides in the PE/PG Phospholipid Membrane, as Determined by FTIR Spectroscopy-FTIR spectroscopy was used to determine the secondary structure of the peptides within the phospholipid membrane. The spectra of the amide I region of the Tar-1 9WT and the Tar-1 9WTϩoct bound to the PE/PG (7:3 w/w) multibilayers are shown in Fig.  5, A and B, respectively. Second derivatives, accompanied by 13-data point Savitsky-Golay smoothing, were calculated to identify the positions of the component bands in the spectra (figure not shown). These wavenumbers were used as initial parameters for curve fitting with Gaussian component peaks. Assignment of the different secondary structures to the various amide I regions was calculated according to the values taken from Jackson and Mantsch (45).
The data revealed different structures for the Tar-1 9WT and 9WTϩoct peptides in PE/PG membranes. The major amide I band of the Tar-1 9WT peptide is located at ϳ1633 cm Ϫ1 (57 Ϯ 3%), corresponding to an aggregated ␤-sheet, whereas that of Tar-1 9WTϩoct has a major peak at 1654 cm Ϫ1 (63 Ϯ 4%), corresponding to an ␣-helical structure. These results imply that conjugation of octanoic acid to the Tar-1 9WT peptide stabilized its helical structure and reduced the aggregation of the peptide.
The effect of the peptides on the multibilayer acyl chain order can be estimated by comparing the CH 2 -stretching dichroic ratio of pure phospholipid multibilayers with that obtained for membranes with bound peptides. The results indicate that incorporation of Tar-1 9WTϩoct into PE/PG membranes had a stronger effect on the order of the membrane than the Tar-1 9WT peptide (data not shown). These results suggest that the Tar-1 9WTϩoct peptide penetrates more efficiently into the hydrophobic core of the membrane than the 9WT peptide.
Structure Determination Using Circular Dichroism (CD) Spectroscopy-We used CD spectroscopy to complement the FTIR structural studies. We characterized the secondary structure of the peptides in a micellar environment (1% SDS). Only the CD spectral profile of Tar-1 9WTϩoct included double minima at ϳ208 and 222 nm, characteristic of an ␣-helical secondary structure (Fig. 6). These results correlate with the FTIR data, which reveal stabilization of an ␣-helical structure for the octylated TM peptide analog.
Detection of CMC-ANS emission was used to indicate the micellization state of the peptide in solution (38). The peptides, at different concentrations (2.5-100 M), were added to water, and the ANS fluorescence was monitored. A dose-dependent increase in the fluorescence signal was observed for the 9WTϩoct peptide (Fig. 7A). However, the 9WT peptide showed a negligible increase in fluorescence, as expected. These results suggest that the CMC of the 9WTϩoct peptide in aqueous solution is ϳ15 M. Higher concentrations will result in peptide micellization.
Visualization of the Global Structure of the Peptides as Determined by Electron Microscopy-Transmission electron microscopy images of Tar-1 9WT and 9WTϩoct peptides were taken to visualize the global structure of the peptide in an aqueous environment at a concentration above the CMC of the 9WTϩoct peptide (Fig. 7, B and C, respectively). Interestingly, the 9WTϩoct peptide self-assembled not to simple micelles but rather to a network of nanofibers. It is possible that attachment of a fatty acid moiety leads to the formation of cylindrical micelles (46).
Statistical Analysis-To determine whether the QXXS motif is a general dimerization motif that can mediate the association of additional TM domains, we analyzed its frequency of occurrence in a broad set of bacterial TM domains. Our results indicate that this motif appears to be significantly over-represented compared with its theoretical expectancy. The motif was found 543 times within a data base of 41,916 TM domains, whereas its average random expected occurrence is 427 Ϯ 22 times (Fig. 8). In comparing the expected values of occurrence of the QXXS and SXXQ motifs, we observed similar distributions, as expected in a sufficiently randomized data base (Fig.  8A). The ratio of occurrence/expected of the QXXS motif is much higher than 1 (Fig. 8B). On the other hand, the ratio of QXXXS approaches 1, suggesting that it is not over-represented in the bacterial data base. A comparison of the QXXS motif with the well studied GXXXG motif revealed a similar ratio of occurrence/expected with p values of 5.5 ϫ 10 Ϫ14 or lower. The difference in the frequency of the GXXXG motif occurrences compared with that of the QXXS motif is probably because of the higher frequency of Gly residues in the membrane as opposed to polar amino acids (47). All together, the results indicate that the QXXS motif may function as a general dimerization motif and therefore appears in additional TM domains.
A closer look at the TM proteins that contain the QXXS motif revealed two interesting observations. First, the QXXS motif appears in membrane proteins with different functions and is not limited to chemotaxis receptors. For example, MCPD_EN-TAE (accession number P21823), from Enterobacter aerogenes, is also a methyl-accepting chemotaxis aspartate transducer like the Tar receptor. FTSW_HELPJ, from Helicobacter pylori (accession number Q9ZJ48), is a membrane protein involved in cell division and is located on the inner membrane. RFBE_ SHIFL, from Shigella flexneri (accession number P37781), is a putative O-antigen transporter. RNFD_VIBCH, from Vibrio cholerae (accession number Q9KT89), may be part of a membrane complex involved in electron transport. These examples demonstrate that the QXXS motif is not restricted by function or species. We also found 108 proteins from pathogenic bacteria, for example, FTSW_HELPJ, RFBE_SHIFL, RNFD_ VIBCH, which are previously listed or HLYB_VIBCH from V. cholerae (accession number P15492; Q9KMU8), which is believed to be responsible for the secretion of hemolysin. If the QXXS motif is indeed involved in the assembly of these proteins, it raises the possibility of a new antimicrobial strategy targeting membrane protein assembly. DISCUSSION The involvement of polar residues in the non-covalent association of TM segments has been previously established (23)(24)(25)(26). Recently, we have demonstrated that the polar motif, QXXS, is involved in the dimerization of the N-terminal TM domain of the E. coli aspartate receptor (Tar-1) (17). In this study we found that QXXS is the minimal motif sufficient for Tar-1 TM-TM homo-and hetero-interactions. Statistical analysis of a bacterial TM data base supports a general role for the QXXS motif in TM assembly. Interestingly, our results also show that it is feasible to control the kinetics of TM peptide release by conjugation to aliphatic acids. This may prove to be a useful tool for modulating receptor activity.
Identifying the QXXS Minimal Dimerization Motif-Identification of the Tar-1 minimal dimerization motif was done by several complementary methods: (i) a homodimerization method, using the ToxR system, which can assess the dimer- ization level of a specific TM domain, (ii) a heterodimerization assay, in which different exogenous peptides are introduced to the ToxR system and their dominant-negative effect on the dimerization of the ToxR chimera is measured, and (iii) statistical analysis of a bacterial TM data base, which can identify the QXXS dimerization motif in the TM domain of other bacterial proteins, thus suggesting that this motif is a common oligomerization motif. Taking into account our previous work (17) and studies by other groups (24 -26) we believe that it is possible to replace the Gln and Ser by other polar residues such as Thr, Glu, Asp, or Asn. However, some combination may not behave in the same manner as the wild type because the specific side chain length and polarity may impose somewhat different structural requirements. Indeed, we found over-representation for other such polar pairs in our TM domain data base. For example, QXXQ occurs 140 times, whereas its expected value is only 86 (ratio of 1.63). We conclude that a Polar-XX-Polar motif can mediate TM dimerization when at least one of the positions is occupied by a strong polar residue (has two polar side-chain atoms as in Glu).
Based on our previous work we estimate that the Gln and Ser contribute about 70 and 30% to the free energy of association, respectively (17). Gratkowski et al. (26) previously estimated the free energy of association for polar amino acids within TM domains. They calculated ⌬⌬G for Gln and Ser to be ϳ1.75 and ϳ0.25 kcal/mole per monomer, respectively. Thus, according to these figures we can estimate the ⌬⌬G of association for Tar-1 as ϳ4 kcal/mol. One major difference that needs to be noted is that the TM domain in their system is a trimer based on the GCN leucine zipper, whereas the Tar-1 is a dimer. Nevertheless, we previously demonstrated that the TM peptide by itself can disturb chemotaxis through the aspartate receptor. Therefore, its energy of association must be significant.
The homodimerization results indicate that a construct containing the C-terminal half of the wild-type Tar-1 TM domain (9TMϩLeu) and a construct containing only the QXXS motif within a leucine and alanine backbone TM domain (QXXS motifϩbackbone) exhibit dimerization levels similar to the wild-type Tar-1 TM domain (Fig. 1). These results suggest that the QXXS motif is the minimal dimerization motif needed for Tar-1 assembly. The heterodimerization results, which showed that the inhibition ability of the 16WT TM peptide and the 9WT-truncated lipopeptide are similar (Fig. 3), further support this conclusion. Previous studies have shown that synthetic TM peptides can disrupt the proper assembly of TM domains (48 -53). However, this is direct evidence that even a short peptide containing a minimal recognition motif is sufficient to inhibit helix-helix interaction of a wild-type homodimer when attached to a fatty acid. These results may also provide a biophysical explanation for a previous study, which demonstrated that a truncated TM peptide can inhibit a T-cell-mediated immune response (53).
Statistical analysis of the occurrence frequency of the QXXS motif suggests a general role for this motif in the assembly of bacterial TM domains. We observed that the QXXS motif is significantly over-represented compared with its theoretical expectancy in a broad set of bacterial TM domains. When comparing this motif to the GXXXG motif, which is known to be over-represented in membrane proteins, we found a similar ratio of occurrence/expected, along with very small p values. The number of occurrences in the data base of the QXXS motif is about 12 times lower than the GXXXG motif. This may be because of the lower frequencies of Gln and Ser compared with that of Gly in the TM domains.
Further experimental studies of other proteins containing the QXXS motif are necessary to establish whether this motif is involved in their assembly. However, there are few examples of other QXXS containing membrane proteins that function as dimers and have two TM domains. Among these are the serine receptor from E. coli (P02942), the serine receptor from E. aerogenes (P21822), the well studied aspartate receptor from Salmonella typhimurium (P02941) (54), and the aspartate receptor from E. aerogenes (P21823).
Identifying the Role of the Surrounding Residues-Using the heterodimerization method, we found that a truncated TM peptide, containing only 9 amino acids out of the Tar-1, exhibit 50% inhibition ability compared with a Tar-1 wild-type peptide (16 amino acids of the Tar-1, Fig. 3). This result can be explained by the exclusion of critical amino acids that are important for the affinity and specificity of the interaction with Tar-1 from the sequence of the peptide. Another possibility is that the hydrophobicity of the peptide is below the threshold needed for proper integration into the membrane, because it contains only half of the TM sequence. To differentiate between these two possibilities, we examined the dominant-negative activity of an identically truncated peptide when conjugated to a fatty acid moiety (9WTϩoct). This peptide regained inhibition levels similar to the wild-type peptide (16WT, Fig. 3). The recovery of the dominantnegative ability of the 9WTϩoct lipopeptide compared with its non-octylated form revealed that the role of the Nterminal portion of the Tar-1 is to confer hydrophobicity and therefore it could be replaced by a fatty acid moiety. According to these results, we speculate that short TM peptides will be able to retain their ability to interact with the wild-type TM domain and therefore to disrupt receptor assembly, providing that the peptides contain the dimerization motif and are able to integrate properly into the membrane. . The occurrence/expectation ratio of QXXS and GXXXG are higher than 1, indicating over-representation in bacterial TM domains. The occurrence/expectation ratio of QXXXS approaches 1, indicating that it is well within the expected distribution; this is also supported by its p value.
Besides compensating for hydrophobicity, it is possible that the octanoic acid, together with the lysine tags, increases the probability of the peptide being inserted into the membrane in its correct orientation and therefore assists the octylated peptide to act as a better inhibitor. In a recent study Freeman-Cook et al. (55) constructed a library in which 15 TM amino acid residues of the E5 protein (44-amino acids) were replaced with random hydrophobic sequences. A glutamine residue, which plays a role both in homodimerization of E5 and in the interaction between the E5 protein and the platelet-derived growth factor ␤ receptor, was preserved. The results of this study revealed that despite remarkable sequence flexibility in other positions, the ability to induce transformation via dimerization of the platelet-derived growth factor ␤ receptor is retained. This further supports our observation that the role of most amino acids in the TM sequence is merely to confer hydrophobicity. These amino acids can be replaced by other hydrophobic residues and still preserve the proper assembly.
Characterization of TM and Lipo-TM Peptides-Structural studies of the two peptides, 9WT and 9WTϩoct, performed by using FTIR (Fig. 5) and CD (Fig. 6) spectroscopy, revealed that conjugation of a lipophilic tail stabilizes an ␣-helical structure. The ␣-helical structure is thermodynamically favored within the low dielectric environment of a biological membrane because it shields polar C ϭ O and NH groups (56). Therefore, increased ␣-helical content of the 9WTϩoct peptide could be induced by inserting the peptide into the membrane environment. The effect of the peptides on the multibilayer acylchain order also indicated that the 9WTϩoct peptide better incorporates into the core of the membrane than the 9WT peptide. SPR studies demonstrated rapid association and dissociation of the 9WT peptide from the PE/PG membrane, whereas the 9WTϩoct exhibited slow kinetics and incomplete release (Fig. 4). This difference in the binding pattern implies that the binding of the 9WT peptide involves a single binding phase, whereas the 9WTϩoct peptide first binds the membrane surface followed by an insertion into the hydrophobic core of the membrane. Taken together, these results support our assumption that the 9WTϩoct peptide inserts into the membrane core, whereas the 9WT peptide is mostly localized to the surface of the membrane. Overall, we concluded that conjugation of a fatty acid moiety stabilizes the ␣-helical structure, compensates for the loss of hydrophobicity, and therefore enables the peptide to be inserted into the membrane core.
Possible Application of Lipo-TM-peptide-Using transmission electron microscopy, we observed the formation of nanofibers by the 9WTϩoct peptide in a concentration above its CMC (Fig. 7). According to existing knowledge, an alkyl tail coupled to an ionic peptide should assemble in water into cylindrical micelles because of the overall conical shape of the amphiphil (46). Formation of cylindrical micellar structures may allow the active peptide to be slowly released and therefore may be applied as a tool for controlled drug release. Note that although the inhibition assays were examined mainly below the CMC, it is possible that some fraction of the lipopeptides aggregated, resulting in a lower peptide concentration than previously assumed. This can imply that the lipopeptides have in fact better inhibition activity than the reported one.
In summary, this study presents a minimal polar dimerization motif that can mediate TM-TM dimerization. A Swiss-Prot survey revealed that this motif is present within the TM domains of pathogenic bacterial proteins. These TM proteins are therapeutically attractive targets for inhibition by short TM peptides. In addition, the results of the inhibition activity of the lipopeptide constitute a novel therapeutic approach whose aim is to inhibit uncontrolled dimerization by using peptides composed of a minimal length.