The interface of a membrane-spanning leucine zipper mapped by asparagine-scanning mutagenesis

Abstract

An oligo-leucine sequence has previously been shown to function as an artificial transmembrane segment that efficiently self-assembles in membranes and in detergent solution. Here, a novel technique, asparagine-scanning mutagenesis, was applied to probe the interface of the self-assembled oligo-leucine domain. This novel approach identifies interfacial residues whose exchange to asparagine leads to enhanced self-interaction of transmembrane helices by interhelical hydrogen bond formation. As analyzed by the ToxR system in membranes, the interface formed by the oligo-leucine domain is based on a leucine-zipper-like heptad repeat pattern of amino acids. In general, the strongest impacts on self-assembly were seen with asparagines located around the center of the sequence, indicating that interaction is be more efficient here than at the termini of the transmembrane domains.

Sequence-specific interactions between α-helical transmembrane segments (TMSs) support folding and assembly of many integral membrane proteins (White and Wimley 1999; Popot and Engelman 2000; Shai 2001; Langosch et al. 2002; DeGrado et al. 2003). These interactions depend on mutual recognition of complementary surfaces of the TM helices. The structures of crystallized membrane proteins revealed that interacting TMSs adopt either positive or negative crossing angles, depending on the geometry of side-chain packing (Bowie 1997; Langosch and Heringa 1998). Positive crossing angles characteristic of left-handed pairs of TM helices result from regular interdigitation of side-chains at a and d positions of an [abcdefg]_n heptad repeat motif, whereas e and g positions are located at the periphery of these helix–helix interfaces (Fig. 1D ▶; Simmerman et al. 1996^; Pinto et al. 1997^; Langosch and Heringa 1998). This heptad pattern was originally identified in soluble leucine zipper interaction domains and gives rise to “knobs-into-holes packing” of side-chains (Lupas 1996). In contrast to that, the interfaces of TMSs crossing at negative angles appear to conform to [abcd]_n repeats, in which a and b correspond to interfacial residues; this is exemplified by self-assembling TMSs from glycophorin A (MacKenzie et al. 1997) or from SNARE proteins (Laage and Langosch 1997; Laage et al. 2000).

Figure 1.

Mapping interfacial residue positions of the oligo-leucine TMS (leu20) by asparagine-scanning mutagenesis in membranes. (A) Self-assembly of leu20 and its asparagine mutants as determined by the ToxR system. Mutants with strongly increased β-galactosidase activities (mean ± SE, n = 12) relative to the parental leu20 sequence follow a heptad repeat pattern and are labeled accordingly. (B) Protein expression levels were comparable as confirmed by Western blot. The order of samples corresponds to that in A. (C) Growth kinetics of PD28 cells expressing the ToxR constructs in minimal medium with maltose were similar, thus indicating similar concentrations of the different ToxR proteins in the membrane. A construct in which the TMS had been deleted (ΔTM) served as negative control. (D) Alignment of leu20 to the heptad pattern revealed by asparagine-scanning mutagenesis. The helical wheel diagram depicts how a and d positions form the interface of a leucine zipper. Although a pair of helices is shown for the sake of simplicity, the same geometry of side-chain packing applies to complexes with more than two helices.

Previously, it has been shown that oligo-leucine sequences form stable α-helices (Zhang et al. 1992) and function as artificial TMSs (Chen and Kendall 1995) that self-interact in membranes and in non-denaturing detergent solution (Gurezka et al. 1999). Self-interaction was preserved with a heptad repeat motif of leucine grafted onto a monomeric oligo-alanine host sequence. Therefore, we suggested that an oligo-leucine helix may be regarded as a minimal model for membrane-spanning leucine zippers (Gurezka et al. 1999). Upon randomization of the heptad motif with different sets of amino acids followed by selection of self-interacting sequences, aliphatic amino-acids were enriched in high-affinity sequences (Gurezka and Langosch 2001).

Because direct proof that an oligo-leucine TMS self-interacts via a heptad-repeat pattern has been lacking, we mapped the residues that are critical for interaction by a novel technique, asparagine-scanning mutagenesis.

Results and Discussion

We expressed a 20-residue oligo-leucine TMS (leu20) and a complete series of asparagine point mutants in the context of chimeric ToxR transcription activator proteins. All proteins were encoded by ToxRIV plasmids (Gurezka and Langosch 2001) in which expression is under control of the inducible arabinose promoter. Upon low-level expression, the single-span ToxR chimeric proteins are anchored within the inner Escherichia coli membrane. There, they self-assemble depending on the mutual affinity of their TMSs as monitored by transcription activation of a β-galactosidase reporter gene (Langosch et al. 1996; Brosig and Langosch 1998). In addition, these proteins were overexpressed, thus forcing them into inclusion bodies that were detergent-solubilized and analyzed by SDS-PAGE.

Rationale of asparagine-scanning mutagenesis

Protein–protein interfaces are frequently mapped by alanine-scanning mutagenesis, as exchange of most residue types to alanine creates voids leading to position-specific reductions of affinity. Because single mutations to alanine reduced self-interaction of an oligo-leucine TMS only slightly (data not shown), we choose an alternative approach. This novel method is based on recent findings that demonstrate that asparagine residues located within TMSs drive their interaction in apolar environments like a lipid membrane or a detergent micelle. This is probably due to formation of strong hydrogen bonds between their side-chains when water molecules are not available as alternative partners for hydrogen bond formation (Choma et al. 2000; Zhou et al. 2000). We therefore reasoned that systematic replacement of the leucine residues by asparagine would result in enhanced TMS–TMS affinity, depending on whether the mutated position was closely juxtaposed to its counterpart within the helix–helix interface or not.

Mapping the oligo-leucine helix–helix interface by asparagine-scanning mutagenesis

We mutated each residue within the oligo-leucine sequence individually to asparagine and first determined the impact of the different mutations on self-interaction in membranes. In contrast to natural TMSs that self-assemble via specific faces localized on one or more sides of the helix (Arkin et al. 1994; Langosch et al. 1996; Simmerman et al. 1996; Pinto et al. 1997; Laage et al. 2000), an oligo-leucine helix is rotation-symmetric with respect to its long axis. Thus, no preferred interface is expected to exist. On the other hand, the orientation of the interacting faces of a TM helix relative to the DNA-binding ToxR domains in the complex has been found to be critical for transcription activation in earlier studies (Langosch et al. 1996; data not shown). Therefore, the functional coupling between the self-interacting oligo-leucine helix and the ToxR domain would be optimal for one interface out of several alternatives. We exploited this spatial constraint to map one of these faces with respect to its underlying residue pattern. Specifically, we asked whether this pattern would correspond to an [abcdefg]_n or to an [abcd]_n pattern.

As shown in Figure 1 ▶, the different mutants exhibited different levels of transcription activation relative to the parental leu20 sequence. Whereas asparagine placed in the N-terminal region of the TMS did not enhance β-galactosidase activities, strong signal increases were exhibited by several mutants between positions 5 and 18. Mutating positions 7, 11, and 14 gave the strongest signals, and position 18, albeit less sensitive than the former positions, responded with a stronger increase than did positions 17 and 19 in its immediate vicinity (Fig. 1A ▶). Thus, residue positions 7, 11, 14, and 18 are likely to be closely juxtaposed in the membrane-embedded TMS–TMS interface. This residue pattern corresponds to the periodicity of a and d positions of the leucine zipper heptad motif (Fig. 1D ▶). Asparagines at most b, c, e, f, and g positions of this central region also elicited significant signal increases. In the framework of the leucine zipper, the distance between these positions would be too large for hydrogen bond formation to occur (Fig. 1D ▶). In these cases, the helices may partially rotate around their long axes to form alternative interfaces in which these asparagines would be juxtaposed. Because rotation would lead to less favorable orientations of the DNA-binding ToxR domains relative to each other in the complex, mutating a, b, c, e, f, and g positions would elicit less transcription activation, which agrees with our observation.

To exclude the possibility that the observed differences in signal strength are due to differences in protein expression and/or efficiencies of membrane integration, we performed two types of control experiments. Western blotting of whole-cell lysates ascertained that the different ToxR proteins were expressed at comparable levels (Fig. 1B ▶). Further, we tested whether their concentrations in inner bacterial membranes was comparable. To this end, we determined the ability of the ToxR proteins to complement the deficiency in maltose-binding protein (MalE) of an E. coli deletion strain (PD28). This strain cannot grow in minimal medium with maltose as the only carbon source unless the C-terminal MalE-domain of ToxR chimeric proteins is successfully translocated to the periplasmic space (Brosig and Langosch 1998). Expression of the asparagine mutants led to cell densities after 72 h that were within 75% of the parental oligo-leucine sequence (Fig. 1C ▶). A cytoplasmically localized construct in which the TMS had been deleted (ΔTM) served as negative control. Thus, the observed degrees of transcription activation are not significantly influenced by the concentrations of the ToxR proteins in the membrane.

Taken together, we conclude that the oligo-leucine helix self-assembles via a leucine zipper type of side-chain packing. The observation that the impact of asparagine on self-assembly was much stronger when placed at the center of the TMS compared with locations at its termini was initially unexpected and will be discussed below.

As outlined above, asparagine-scanning mutagenesis is thought to map one out of several potential interfaces of the rotation-symmetric oligo-leucine helix due to spatial constraints of the ToxR system. On the other hand, asparagines at any positions are expected to drive helix–helix interaction in the absence of these constraints. Accordingly, we monitored oligomer formation directly in detergent solution. The ToxR proteins harboring leu20 or its asparagine mutants were overexpressed in E. coli, which forces them into inclusion bodies that were solubilized with the detergent [(3-cholami-dopropyl)dimethylammonio]-1-propane-sulfonate (CHAPS). SDS-PAGE analysis of an unboiled sample of the leu20 protein revealed the presence of multiple high-molecular-weight bands indicative of homomeric adducts (Fig. 2 ▶, left) that are thought to form by leucine–leucine interaction. These adducts corresponded to molecular weights from ~140 to 300 kD that may correspond to homodimers and higher homooligomers of the 70-kD monomeric species. Boiling of the sample prior to SDS-PAGE dissociated the leu20 oligomers. However, mutants L8N through L15N formed stable adducts that were seen even after sample boiling (Fig. 2 ▶, right). These adducts were similar in size to the ones seen with the unboiled leu20 sample, although the relative strengths of the adduct bands varied somewhat with the type of mutant. Thus, asparagine residues appear to stabilize helix–helix interactions but do not radically change the stoichiometries of the oligomeric species. Importantly, all mutations between positions 8 and 15 promoted oligomerization with comparable efficiencies, whereas neither N- nor C-terminal regions responded to asparagine mutation with significant oligomer formation. In isotropic detergent solution, therefore, it appears that asparagine anywhere within the central region orients the helices such as to allow for hydrogen bond formation and self-assembly.

Figure 2.

Self-assembly in detergent solution as analyzed by SDS-PAGE analysis. Proteins overexpressed in TOP10 cells were solubilized, electrophoresed at equivalent starting concentrations in sample buffer, and visualized by Western blotting. (Left) Electrophoresis of the leu20 protein not boiled in SDS sample buffer. (Right) leu20 was monomeric upon sample boiling, whereas its asparagine mutants still formed various oligomeric species when asparagine was located between positions 8 and 15.

What is the reason for the stronger impact of asparagine at the center of the helix compared with at its termini? In a related study (Ruan et al. 2004), we encountered a similar situation when we mapped the interface between self-associating erythropoietin receptor TMSs by asparagine-scanning mutagenesis. There, we initially asked whether the smaller impact of asparagine at the N-terminal positions is due to steric hindrance of TMS–TMS interaction by the adjacent ToxR domains. Because, however, insertion of a flexible linker sequence between ToxR domain and TMS did not influence the distribution of the impact of asparagine, we excluded this possibility. Another possible explanation for our observation is that oligo-leucine helices are not flexible enough to allow for formation of an interface extending over more than two or three helical turns. In this case, asparagines within the terminal regions would be too far apart from their counterparts in the neighboring helices for hydrogen bonding to occur. Finally, promotion of TMS–TMS assembly by hydrogen bonds may depend on the latter’s position in the lipid bilayer. Because hydrogen bonds are essentially electrostatic in nature, their strength critically depends on the polarity of the immediate environment. Consequently, a polarity gradient across the membrane may influence the impact of asparagine on TMS–TMS assembly. Indeed, electron paramagnetic resonance spectroscopy previously revealed that the polarity of model membranes composed of mono-unsaturated phospholipids increased from the middle of the acyl chain region toward the head group regions (Subczynski et al. 1994). The E. coli inner membrane containing approximately equal fractions of saturated and mono-unsaturated acyl chains (Neidhardt et al. 1996) may exhibit a similar polarity gradient. Accordingly, TMSs may experience the lowest polarity at their central regions; this, in turn, could account for increased strength of hydrogen bonding between asparagine residues at a central position, which is consistent with our findings. The same argument would apply to TMSs in SDS-micelles because their terminal regions would be close to the charged surface of the micelle, whereas the central regions would be embedded in its strongly hydrophobic core. This conclusion is supported by a recent study (Lear et al. 2003) addressing the effect of asparagine on self-assembly of a designed hydrophobic leucine zipper. By analytical ultracentrifugation of corresponding synthetic peptides in detergent micelles, these investigators showed that asparagine residues within the apolar region of the peptide provide a significantly larger driving force on helix–helix interaction than an asparagine near the apolar/polar interface of the micelle.

Conclusions

Our present results reveal the nature of the interface formed when an oligo-leucine helix self-assembles to noncovalent complexes. These complexes are strongly stabilized upon introduction of asparagine residues at certain positions. When analyzed in the context of the ToxR system, the most sensitive positions correspond to a heptad repeat pattern of residues, indicating a leucine zipper type of side-chain packing. Further, asparagines had a much stronger impact on self-assembly when located at the center of the oligo-leucine sequence than at the termini. Although the mechanism underlying the latter observation is presently not entirely clear, we believe that it is related to polarity gradients in lipid bilayers and detergent micelles that modulate the strength of hydrogen bonds between the asparagine residues. This might have far-reaching implications for the structure and function of membrane proteins. The major forces driving interactions of natural TMSs, that is, van der Waals forces and weak hydrogen bonds between polar side-chains, C_α hydrogen atoms. and the peptide backbone (Popot and Engelman 2000; Senes et al. 2001; Langosch et al. 2002), are all mainly electrostatic in nature, and their strengths are therefore inversely related to the polarity of the solvent (Israelachvili 1991; Daniel et al. 2002). It would follow that TMS–TMS interactions are generally stronger in the low-polarity center of lipid bilayers than at sites closer to the head group regions. Conceivably, such gradients of binding affinity could have implications for signaling mechanisms across lipid membranes.

Methods and materials

Plasmid constructs

Construction of plasmid pToxR IVΔTM was described previously (Gurezka and Langosch 2001). Plasmids pToxRIV leu20 was constructed by ligating a synthetic oligonucleotide cassette encoding the desired sequence into pToxRIV previously cut with NheI and BamHI. Each point-mutant was constructed by the Kunkel method (Kunkel et al. 1987) by using the Biorad T7 mutagenesis kit according to manufacturer’s instructions. All constructs were verified by dideoxy sequencing.

ToxR activity assays and control experiments

Plasmid-transformed E. coli FHK12 cells were grown in Luria-Bertani medium (LB) for 24 h at 37°C under shaking in the presence of 2% (w/v) glucose, 0.08% (w/v) l-arabinose, 0.4 mM isopropyl-1-thio-β-d-galactopyranoside, and 33 μg/mL kanamycin, (Gurezka and Langosch 2001). β-Galactosidase activity of cell-free extracts was determined as described previously (Langosch et al. 1996) and is given in Miller units (MU ± SE, n = number of data points). Western blotting was done as previously (Langosch et al. 1996) by using a MalE antiserum (New England Biolabs). Functional complementation of MalE deficiency of PD28 cells was done as described (Brosig and Langosch 1998) by measuring growth kinetics of expressing cells in minimal medium with maltose as the only carbon source.

Protein overexpression

Plasmid-transformed E. coli TOP10 cells were grown in LB until OD₆₀₀ = 0.5 to 0.6. Arabinose was added to a final concentration of 0.001%, and the culture was grown for 3 h at 37°C with constant shaking. Cells were pelleted and resuspended into 50 mM HEPES (pH 7.9), 5 mM EDTA, 1 mM PMSF, 0.025% NaN₃, and 0.1 mg/mL lysozyme. After three freeze/thaw steps in liquid nitrogen, cells were incubated in the presence of 10 μg/mL DNase I for 15 min at room temperature followed by centrifugation at 10,000g for 15 min at 4°C. The pellet was resuspended into solubilization buffer containing 25 mM HEPES (pH 7.9), 500 mM NaCl, 2% (w/v) CHAPS, and 1 mM EDTA and shaken for 45 min at 4°C. After centrifugation at 100,000g for 20 min at 4°C, the supernatant that contains the overexpressed proteins was collected for SDS analysis.

SDS-PAGE

Samples containing 4% (w/v) SDS were electrophoresed in 5% to 12.5% polyacrylamide gradient gels. Gels with unboiled samples were run at 4°C. Western blots were developed with an antibody against maltose binding protein (New England Biolabs).