Transmembrane glycine zippers: physiological and pathological roles in membrane proteins

Abstract

We have observed a common sequence motif in membrane proteins, which we call a glycine zipper. Glycine zipper motifs are strongly overrepresented and conserved in membrane protein sequences, and mutations in glycine zipper motifs are deleterious to function in many cases. The glycine zipper has a significant structural impact, engendering a strong driving force for right-handed packing against a neighboring helix. Thus, the presence of a glycine zipper motif leads directly to testable structural hypotheses, particularly for a subclass of glycine zipper proteins that form channels. For example, we suggest that the membrane pores formed by the amyloid-beta peptide in vitro are constructed by glycine zipper packing and find that mutations in the glycine zipper motif block channel formation. Our findings highlight an important structural motif in a wide variety of normal and pathological processes.

Fig. 1.

Homooligomeric glycine zipper channel structures. Shown are KcsA, potassium channel pore-lining helices (Protein Data Bank ID code 1BL8) (11); MscL, mechanosensitive channel of large conductance (PDB ID code 1MSL) (12); VacA, vacuolating toxin A anion selective channel model (PDB ID code 1SEW) (13); and MscS, mechanosensitive channel of small conductance (PDB ID code 1MXM) (14). The balls highlight the C^α carbon positions of the glycine zipper packing residues. The glycine zipper packing residues are also highlighted on the TM sequences shown below. Small residues (Gly, Ala, or Thr) were overrepresented at the highlighted positions on the homologs.

Fig. 2.

Amino acid preferences and conservation of glycine zippers. (A) Glys in glycine zipper motifs are more conserved than random Gly residues. A histogram of the RCS for a set of 19 glycine zipper motif proteins is plotted. A ratio >1 indicates that Gly residues are more conserved in the glycine zipper motif than they are elsewhere in the structure. Conservation scores and protein codes are given in Table 2. (B) Odds ratios for Ala, Ser, and Thr substitutions. (C) Comparison of odds ratios for the strong glycine zipper motifs with single substitutions of Val, Leu, or Ile. Many triplet motifs with the small residues (Gly, Ala, Ser, or Thr) spaced four apart are at least somewhat overrepresented in TM helices. Single substitutions with Val, Leu, or Ile are not overrepresented, indicating that they are not generally good substitutes for Gly. (D) A glycine zipper drives a preference for right-handed helix packings. Glycine residues located at every fourth position, represented as balls on cylindrical model helices, create a polar stripe of small residues by exposing the backbone carbonyl and amide atoms that may nucleate helix interactions in the apolar membrane environments. The right-handed packing found in the glycine zipper helices of aquaporin 1 (PDB ID code 1J4N, residues 212-231), glycerol-3-phospate transporter (PDB ID code 1PW4, residues 380-404), and ABC transporter (PDB ID code 1L7V, residues 92-107) are highlighted.

Fig. 3.

Channel formation of Aβ42 WT peptide and mutants. (A) Sequences of the Aβ42 peptide variants. The position of the mutation in the glycine zipper motif is indicated by the arrow. (B) Channel activities of Aβ42 peptides recorded in the symmetrical K⁺ solutions (140 mM KCl/1 mM CaCl/10 mM Hepes, pH7.4) at a TM potential of -200mV (see Methods for details). Multiple conductance levels were found with the WT Aβ42 peptides, indicating the presence of multiple channels in the membrane. (C) The current-voltage relationships for the WT Aβ42 peptides. A single channel conductance of ≈12 pS was calculated from the difference in slopes of the regression line drawn to fit the Aβ42 channel current (▪, 24.1 pS) and the baseline current from the lipid bilayer (♦, 11.9 pS).

Fig. 4.

Neuronal cell viability after treatment of Aβ42 WT and mutants. The percentage of dead Neuro-2a cells was measured after exposure of the culture to 5 μMAβ42 peptide for 5 h. With the WT Aβ42 peptide, 37.8 ± 7.0% of the cells were dead after 5 h. With G29L, G33L, and G37L, only 21.5 ± 2.4%, 9.4 ± 4.3%, and 3.7 ± 1.7% were dead, respectively. Thus, all of the mutants were less toxic than WT. The order of increasing toxicity was WT > G29L > G33L > G37L. This order correlates well with the mutants' ability to disrupt membranes, suggesting that toxicity in vitro may depended on membrane permeabilization.