By a number of measures, alanine is poised at the threshold between those amino acids that promote the membrane integration of transmembrane ␣-helices and those that do not. We have measured the preference of alanine to partition into the lipid-water interface region over the central acyl chain region of the endoplasmic reticulum (ER) membrane both by its ability to promote the formation of so-called helical hairpins, i.e. a pair of transmembrane helices separated by a tight turn, and by mapping the position relative to the membrane of the lumenal end of a transmembrane ␣-helix that ends with a block of 10 alanines. Both measures show that Ala has a weak but distinct preference for the interface region, which is in agreement with recent biophysical measurements on pentaeptide partitioning in simple water-lipid or water-octanol systems (Jayasinghe, S., Hristova, K., and White, S. H. (2001) J. Mol. Biol. 312, 927-934). Considering the complexity of the translocon-mediated insertion of membrane proteins into the ER, the agreement between the biochemical and biophysical measurements is striking and suggests that proteinlipid interactions are already important during the very early steps of membrane protein assembly in the ER.Most integral membrane proteins are composed of tightly packed bundles of transmembrane ␣-helices (1). The loops connecting the helices tend to be short (2), suggesting the possibility that pairs of helices with a short connecting loop, i.e. a helical hairpin (3), may be inserted en bloc into the membrane (4 -6) rather than the helices being recognized and inserted one by one by the translocation/membrane insertion machinery (7).Using in vitro translation in the presence of rough microsomes (RMs) 1 of model membrane proteins containing engineered poly-Leu segments of various lengths, we have shown previously that the balance between forming a long, continuous transmembrane helix and a helical hairpin can be a delicate one; in constructs with 35-40-residue-long poly-Leu stretches, a single Leu 3 Pro mutation near the middle of the poly-Leu stretch is sufficient to convert a long transmembrane helix into a helical hairpin (8, 9). Other polar and charged residues also promote helical hairpin formation in this context (9 -11). As an example of a similar effect in a wild type protein, an Asn-Pro 3 Leu-Leu mutation between two tightly spaced transmembrane helices in the Saccharomyces cerevisiae Ste14p protein has been shown to convert the helical hairpin into a single, long transmembrane helix (12).In our previous studies (9, 10), Ala has not shown any tendency to promote the formation of helical hairpins. The question of whether Ala should be considered a "membrane-seeking" or a "water-seeking" residue is not an easy one, however, as Ala balances close to the hydrophobic/hydrophilic threshold in many well known hydrophobicity scales (13). Poly-Ala segments of ϳ20 residues or more can insert as transmembrane helices during membrane protein assembly into the endoplasmic reticulum (ER) membrane (14), bu...