Saturation site-directed mutagenesis of thymidylate synthase

We have subjected 12 different codons of a synthetic Lactobacillus casei thymidylate synthase (TS) gene to saturation site-directed mutagenesis to create amino acid "replacement sets" at each of those positions. The target residues were chosen because they are highly conserved and because they are important for the structure and function of the protein as indicated by solution and structural studies. The mutagenesis procedure involved excision of a fragment of the synthetic gene containing the target codon, followed by its replacement with a mixture of oligonucleotides which code for all 20 amino acids and the amber stop codon. TS mutants were identified by DNA sequencing, and catalytically active mutants were identified by genetic complementation using a Thy- strain of Escherichia coli. Only 3 of the 12 target amino acids examined were essential for TS activity; and of the 125 total mutants identified, 57 were catalytically active. These results point to a high degree of plasticity of TS in accommodating function with structural change.     

Molecular contacts in the transmembrane c-subunit oligomer of F-ATPases identified by tryptophan substitution mutagenesis

When isolated in its monomeric form, subunit c of the proton transporting ATP synthase of Escherichia coli was shown to fold in a hairpin-like structure consisting of two hydrophobic membrane spanning helices and a short connecting hydrophilic loop. In the plasma membrane of Escherichia coli, however, about 9-12 c-subunit monomers form an oligomeric complex that functions in transmembrane proton conduction and in energy transduction to the catalytic F1 domain. The arrangement of the monomers and the molecular architecture of the complex were studied by tryptophan scanning mutagenesis and restrained MD simulations. Residues 12-24 of the N-terminal transmembrane segment of subunit c were individually substituted by the large and moderately hydrophobic tryptophan side chain. Effects on the activity of the mutant proteins were studied in selective growth experiments and various ATP synthase specific activity assays. The results identify potential intersubunit contacts and structurally non-distorted, accessible residues in the c-oligomer and add constraints to the arrangement of monomers in the oligomeric complex. Results from our mutagenesis experiments were interpreted in structural models of the c-oligomer that have been obtained by restrained MD simulations. Different stoichiometries and monomer orientations were applied in these calculations. A cylindrical complex consisting of 10 monomers that are arranged in two concentric rings with the N-terminal helices of the monomers located at the periphery shows the best match with the experimental data.     

Random mutagenesis targeted to the psbAII gene of Synechocystis sp. PCC 6803 to identify functionally important residues in the D1 protein of the photosystem II reaction center

More than one hundred mutants of Synechocystis sp. PCC 6803 impaired in photoautotrophic growth were generated by in vitro random PCR mutagenesis targeted to a region of the psbAII gene corresponding to a 210 amino acid (Ser148-Ala357) segment of the D1 protein. The 90 random mutants that could translate the full-length D1 protein carried 1-9 (on average 3.0) amino acid substitutions in the targeted region. Mutations were often found in the obligate photoheterotrophic strains at specific residues that have been reported or speculated to be important in the function of PSII, such as Y161, H198, H272, E333 and H337. This verifies the usefulness of the present method to identify functionally important residues in PSII. Other residues that were often mutated in the strains with impaired photoautotrophy included non-charged residues around the lumenal edges of transmembrane helices C, D and E, such as I192 and N296. Eleven mutants carried a single-point mutation in residues, such as Q165, Q187, W278, A294 and N298, and these identified the functional importance of these residues, most of which were on the donor side of PSII. A preliminary characterization of some of the mutants obtained in this study is provided.     

Improved peptide function from random mutagenesis over short 'windows'

We have applied random mutagenesis over short contiguous residue tracts ('windows') within an active peptide (the alpha-peptide of beta-galactosidase) such that all window residues are replaced simultaneously. A novel technique using mixed synthetic oligonucleotides and selection against an EcoK restriction site has allowed the construction of libraries of mutants for two separate windows, sites A and B. Mutant phenotypes can be easily assessed in vivo by a complementation test, and panels of mutants have been quantitatively tested in vitro. This allowed the rapid probing of structural requirements for each site. The two windows yielded markedly disparate results. Site B was much less stringent in its sequence requirements for significant function than Site A, and mutants with improved function were isolated at Site B alone. In addition, one Site B mutant with wild-type levels of activity showed enhanced stability to heat or a protein denaturant. We propose that short tracts with the characteristics of Site B constitute 'secondary' interaction sites which are more tolerant of sequence diversity. Random manipulation of such secondary sites is thus more likely to yield upmutations for standard or altered environments. Window mutagenesis can in principle be applied to any protein--protein or protein--ligand interaction.     

Tryptophan-216 is essential for the transglycosylation activity of endo-beta-N-acetylglucosaminidase A

Endo-beta-N-acetylglucosaminidase from Arthrobacter protophormiae (Endo-A) has a high level of transglycosylation activity. To determine which amino acids are involved in this activity, we employed deletion analysis, as well as random and site-directed mutagenesis. Using PCR random mutagenesis, 11 mutants with greatly decreased levels of enzyme activity were isolated. Six catalytically essential amino acids were identified by site-directed mutagenesis. Mutants E173G, E175Q, D206G, and D270N had markedly reduced hydrolysis activity, while mutants V109D, E173D, and E173Q lost all enzymatic activity, indicating that Val-109 and Glu-173 are important for the catalytic function. Moreover, we isolated a random mutation that abolished the transglycosylation activity without affecting the hydrolysis activity. The Trp-216 to Arg mutation was identified, by site-directed mutagenesis, as that responsible for the loss of transglycosylation activity. While other mutants of Trp-216 showed reduced activity, mutation to another positively charged residue (Lys) also abolished the transglycosylation activity. Sequence comparison with two other endo-beta-N-acetylglucosaminidases, that possess transglycosylation activity and that have been cloned recently, reveals a high degree of identity in the N-terminal regions of the three enzymes. These results indicate that the tryptophan residue at position 216 of Endo-A has a key role in the transglycosylation.     

Engineering a large protein by combined rational and random approaches: stabilizing the Clostridium thermocellum cellobiose phosphorylase

The Clostridium thermocellum cellobiose phosphorylase (CtCBP) is a large protein consisting of 812 amino acids and has great potential in the production of sugar phosphates, novel glycosides, and biofuels. It is relatively stable at 50 °C, but is rapidly inactivated at 70 °C. To stabilize CtCBP at elevated temperatures, two protein-engineering approaches were applied, i.e. site-directed mutagenesis based on structure-guided homology analysis and random mutagenesis at various mutation rates. The former chose substitutions by comparison of the protein sequences of CBP homologs, utilized structural information to identify key amino acid residues responsible for enhanced stability, and then created a few variants accurately. The latter constructed large libraries of random mutants at different mutagenesis frequencies. A novel combinational selection/screening strategy was employed to quickly isolate thermostability-enhanced and active variants. Several stability-enhanced mutants were obtained by both methods. Manually combining the stabilizing mutations identified from both rational and random approaches led to the best mutant (CM3) with the halftime of inactivation at 70 °C extended from 8.3 to 24.6 min. The temperature optimum of CM3 was increased from 60 to 80 °C. These results suggested that a combination of rational design and random mutagenesis could have a solid basis for engineering large proteins.     

Mutational analysis of the active site of indoleglycerol phosphate synthase from Escherichia coli.

Indoleglycerol phosphate synthase catalyzes the ring closure of 1-(2-carboxyphenylamino)-1-deoxyribulose 5''-phosphate to indoleglycerol phosphate, the fifth step in the pathway of tryptophan biosynthesis from chorismate. Because chemical synthesis of indole derivatives from arylamino ketones requires drastic solvent conditions, it is interesting by what mechanism the enzyme catalyzes the same condensation reaction. Seven invariant polar residues in the active site of the enzyme from Escherichia coli have been mutated directly or randomly, to identify the catalytically essential ones. A strain of E. coli suitable for selecting and classifying active mutants by functional complementation was constructed by precise deletion of the trpC gene from the genome. Judged by growth rates of transformants on selective media, mutants with either S58 or S60 replaced by alanine were indistinguishable from the wild-type, but R186 replaced by alanine was still partially active. Saturation random mutagenesis of individual codons showed that E53 was partially replaceable by aspartate and cysteine, whereas K114, E163, and N184 could not be replaced by any other residue. Partially active mutant proteins were purified and their steady-state kinetic and inhibitor binding constants determined. Their relative catalytic efficiencies paralleled their relative complementation efficiencies. These results are compatible with the location of the essential residues in the active site of the enzyme and support a chemically plausible catalytic mechanism. It involves two enzyme-bound intermediates and general acid-base catalysis by K114 and E163 with the support of E53 and N184.     

Mutagenesis of a plastidial lysophosphatidic acid acyltransferase

A combination of site-directed and random mutagenesis generated sequence variants of a plastidial lysophosphatidic acid acyltransferase. Alanine substitutions of residues present within two conserved motifs including the putative catalytic histidine resulted in a loss of acyltransferase activity assessed as complementation competence. Substitutions at five sites within the central core resulted in reduced or loss of activity. Truncation mutants reveal that sequences in the C-terminal moiety are essential for function.     

Design and Utility of Temperature-Sensitive Gal4 Mutants for Conditional Gene Expression in Drosophila

Temperature-sensitive (ts) mutants are valuable tools to study the function of essential genes in vivo. Despite their widespread use, little is known about mechanisms responsible for the temperature-sensitive (ts) phenotype, or of the transferability of ts mutants of a specific gene between organisms. Since ts mutants are typically generated by random mutagenesis it is difficult to isolate such mutants without efficient screening procedures. We have recently shown that it is possible to obtain ts mutants at high frequency by targeted mutations at either predicted, buried residues important for protein stability or at functional, ligand binding residues. The former class of residues can be identified solely from amino acid sequence and the latter from Ala scanning mutagenesis or from a structure of the protein:ligand complex. Several ts mutants of Gal4 in yeast were generated by mutating both categories of residues. Two of these ts mutants were also shown to result in tight and rapid ts reporter gene-expression in Drosophila when driven by either the elav or GMR promoters. We suggest possible mechanisms that might be responsible for such transferable ts phenotypes and also discuss some of the limitations and difficulties involved in rational design of ts mutants.     

Design and utility of temperature-sensitive Gal4 mutants for conditional gene expression in Drosophila

Temperature-sensitive (ts) mutants are valuable tools to study the function of essential genes in vivo. Despite their widespread use, little is known about mechanisms responsible for the temperature-sensitive (ts) phenotype, or of the transferability of ts mutants of a specific gene between organisms. Since ts mutants are typically generated by random mutagenesis it is difficult to isolate such mutants without efficient screening procedures. We have recently shown that it is possible to obtain ts mutants at high frequency by targeted mutations at either predicted, buried residues important for protein stability or at functional, ligand binding residues. The former class of residues can be identified solely from amino acid sequence and the latter from Ala scanning mutagenesis or from a structure of the protein:ligand complex. Several ts mutants of Gal4 in yeast were generated by mutating both categories of residues. Two of these ts mutants were also shown to result in tight and rapid ts reporter gene-expression in Drosophila when driven by either the elav or GMR promoters. We suggest possible mechanisms that might be responsible for such transferable ts phenotypes and also discuss some of the limitations and difficulties involved in rational design of ts mutants.     

Structural organization of the N-terminal domain of apolipoprotein A-I: studies of tryptophan mutants

Site-directed mutagenesis and detailed fluorescence studies were used to study the structure and dynamics of recombinant human proapolipoprotein (proapo) A-I in the lipid free state and in reconstituted high-density lipoprotein (rHDL) particles. Five different mutants of proapoA-I, each containing a single tryptophan residue, were produced in bacteria corresponding to each of the naturally occurring Trp residues (position -3 in the pro-segment, 8, 50, 72, and 108) in the N-terminal half of the protein. Structural analyses indicated that the conservative Phe-Trp substitutions did not perturb the conformation of the mutants with respect to the wild-type protein. Steady-state fluorescence studies indicated that all of the Trp residues exist in nonpolar environments that are highly protected from solvent in both the lipid-free and lipid-bound forms. Time-resolved lifetime and anisotropy studies indicated that the shape of the monomeric form of proapoA-I is a prolate ellipsoid with an axial ratio of about 6:1. In addition, the region surrounding Trp 108 appears to be more mobile than the rest of the protein in the lipid-free state. However, in rHDL particles, no significant domain motion was detected for any of the Trp residues. The results presented in this work are consistent with a model for monomeric lipid-free proapoA-I in which the N-terminal half of the molecule is organized into a bundle of helices.     

Enzymatic properties of rat DNA polymerase beta mutants obtained by randomized mutagenesis

We have used random sequence mutagenesis to generate mutants of DNA polymerase β in an effort to identify amino acid residues important for function, catalytic efficiency and fidelity of replication. A library containing 100 000 mutants at residues 274–278 in the N-helix of the thumb subdomain of the polymerase was constructed and screened for polymerase activity by genetic complementation. The genetic screen identified 4000 active pol β mutants, 146 of which were sequenced. Each of the five positions mutagenized tolerated substitutions, but residues G274 and F278 were only found substituted in combination with mutations at other positions. The least conserved residue, D276, was replaced by a variety of amino acids and, therefore, does not appear to be essential for function. Steady-state kinetic analysis, however, demonstrated that D276 may be important for catalytic efficiency. Mutant D276E exhibited a 25-fold increase in catalytic efficiency over the wild-type enzyme but also a 25-fold increase in G:T misincorporation efficiency. We present a structural model that can account for the observations and we discuss the implications of this study for the question of enzyme optimization by natural selection.     

Metal-induced changes in the fluorescence properties of tyrosine and tryptophan site-specific mutants of oncomodulin.

Oncomodulin is a 108-residue, oncodevelopmental protein containing two calcium-binding sites identified as the CD- and EF-loops. The protein contains no tryptophan and only two tyrosine residues, one which is a calcium ligand in the CD-loop (Tyr-57) and one which lies in the flanking D-helix of this loop (Tyr-65). Site-specific mutagenesis was performed to yield five mutants, two with phenylalanine substituted for tyrosine in positions 57 and 65 and three with tryptophan substituted into position 57 in the CD-loop, position 65 in the D-helix, and position 96 in the EF-loop. The single Tyr-containing mutants demonstrated that position 57 was perturbed to a significantly greater extent than position 65 upon calcium binding. Although both tyrosine residues responded to decalcification, the fluorescence intensity changes were in opposite directions, with the more dominant Tyr-57 accounting for the majority of the intrinsic fluorescence observed in native oncomodulin. The substitution of tryptophan for each tyrosyl residue revealed that in both positions the tryptophan resided in polar, conformationally heterogeneous environments. The environment of Trp-57 was affected by Ca2+ binding to a much greater extent compared to that of Trp-65. Only 1 equiv of Ca2+ was required to produce greater than 70% of the Trp fluorescence changes in positions 57 and 65, indicating that Ca2+ binding to the higher affinity EF-loop had a pronounced effect on the protein structure.(ABSTRACT TRUNCATED AT 250 WORDS)     

Introduction of reactive cysteine residues in the epsilon subunit of Escherichia coli F1 ATPase, modification of these sites with tetrafluorophenyl azide-maleimides, and examination of changes in the binding of the epsilon subunit when different nucleotides are in catalytic sites.

Cysteine residues have been exchanged for serine residues at positions 10 and 108 in the epsilon subunit of the Escherichia coli F1 ATPase by site-directed mutagenesis to create two mutants, epsilon-S10C and epsilon-S108C. These two mutants and wild-type enzyme were reacted with [14C]N-ethylmaleimide (NEM) to examine the solvent accessibility of Cys residues and with novel photoactivated cross-linkers, tetrafluorophenyl azide-maleimides (TFPAM's), to examine near-neighbor relationships of subunits. In native wild-type F1 ATPase, NEM reacted with alpha subunits at a maximal level of 1 mol/mol of enzyme (1 mol/3 alpha subunits) and with the delta subunit at 1 mol/mol of enzyme; other subunits were not labeled by the reagent. In the mutants epsilon-S10C and epsilon-S108C, Cys10 and Cys108, respectively, were also labeled by NEM, indicating that these are surface residues. Reaction of wild-type enzyme with TFPAM's gave cross-linking of the delta subunit to both alpha and beta subunits. Reaction of the mutants with TFPAM's also cross-linked delta to alpha and beta and in addition formed covalent links between Cys10 of the epsilon subunit and the gamma subunit and between Cys108 of the epsilon subunit and the alpha subunit. The yield of cross-linking between sites on epsilon and other subunits depended on the nucleotide conditions used; this was not the case for delta-alpha or delta-beta cross-linked products. In the presence of ATP+EDTA the yield of cross-linking between epsilon-Cys10 and gamma was high (close to 50%) while the yield of epsilon-Cys108 and alpha was low (around 10%).(ABSTRACT TRUNCATED AT 250 WORDS)     

Domain complementation studies reveal residues critical for the activity of the mannitol permease from Escherichia coli

This paper presents domain complementation studies in the mannitol transporter, EII^mtl, from Escherichia coli. EII^mtl is responsible for the transport and concomitant phosphorylation of mannitol over the cytoplasmic membrane. By using tryptophan-less EII^mtl as a basis, each of the four phenylalanines located in the cytoplasmic loop between putative transmembrane helices II and III in the membrane-embedded C domain were replaced by tryptophan, yielding the mutants W97, W114, W126, and W133. Except for W97, these single-tryptophan mutants exhibited a high, wild-type-like, binding affinity for mannitol. Of the four mutants, only W114 showed a high mannitol phosphorylation activity. EII^mtl is functional as a dimer and the effect of these mutations on the oligomeric activity was investigated via heterodimer formation (C/C domain complementation studies). The low phosphorylation activities of W126 and W133 could be increased 7-28 fold by forming heterodimers with either the C domain of W97 (IIC^mtlW97) or the inactive EII^mtl mutant G196D. W126 and W133, on the other hand, did not complement each other. This study points towards a role of positions 97, 126 and 133 in the oligomeric activation of EII^mtl. The involvement of specific residue positions in the oligomeric functioning of a sugar-translocating EII protein has not been presented before.     

Structural basis of anthrax edema factor neutralization by a neutralizing antibody

Fine epitope mapping of EF13D, a highly potent neutralizing monoclonal antibody specific for the anthrax edema factor (EF), was accomplished through random mutagenesis and yeast surface display. A yeast-displayed library of single point mutants of an EF domain III (DIII), comprising amino acids 624-800, was constructed by random mutagenesis and screened for reduced binding to EF13D. With this method, residues Leu 667, Ser 668, Arg 671, and Arg 672 were identified as key residues important for EF13D binding. They form a contiguous patch on a solvent-exposed surface at one end of the four-helix bundle of DIII. Computational protein-protein docking experiments between anEF13D model and a crystal structure of EF indicate that the EF13D heavy chain complementarity-determining region 3 (HCDR3) is deeply buried within a hydrophobic cleft between two helices of DIII and interacts directly with residues Leu 667, Ser 668, Arg 671 and Arg 672, providing an explanation for the high binding affinity. In addition, they show that the HCDR3 binding site overlaps with the binding site of the N-terminal lobe of calmodulin (CaM), an EF enzymatic activator, consistent with a previous finding showing direct competition with CaM that results in neutralization of EF. Identifying the neutralization epitope of EF13D on EF improves our understanding of the neutralization mechanism and has implications for vaccine development.     

Transmembrane domains of hepatitis C virus envelope glycoproteins: residues involved in E1E2 heterodimerization and involvement of these domains in virus entry

The transmembrane (TM) domains of hepatitis C virus (HCV) envelope glycoproteins E1 and E2 have been shown to play multiple roles during the biogenesis of the E1E2 heterodimer. By using alanine scanning insertion mutagenesis within the TM domains of HCV envelope glycoproteins, we have previously shown that the central regions of these domains as well as the N-terminal part of the TM domain of E1 are involved in heterodimerization. Here, we used a tryptophan replacement scan of these regions to identify individual residues that participate in those interactions. Our mutagenesis study identified at least four residues involved in heterodimerization: Gly 354, Gly 358, Lys 370, and Asp 728. Interestingly, Gly 354 and Gly 358 belong to a GXXXG oligomerization motif. Our tryptophan mutants were also used to generate retrovirus-based, HCV-pseudotyped particles (HCVpp) in order to analyze the effects of these mutations on virus entry. Surprisingly, two mutants consistently displayed higher infectivity compared to that of the wild type. In contrast, HCVpp infectivity was strongly affected for many mutants, despite normal E1E2 heterodimerization and normal levels of incorporation of HCV glycoproteins into HCVpp. The characterization of some of these HCVpp mutants in the recently developed in vitro fusion assay using fluorescent-labeled liposomes indicated that mutations reducing HCVpp infectivity without altering E1E2 heterodimerization affected the fusion properties of HCV envelope glycoproteins. In conclusion, this mutational analysis identified residues involved in E1E2 heterodimerization and revealed that the TM domains of HCV envelope glycoproteins play a major role in the fusion properties of these proteins.     

Redesigning the substrate specificity of an enzyme: isocitrate dehydrogenase

Despite the structural similarities between isocitrate and isopropylmalate, isocitrate dehydrogenase (IDH) exhibits a strong preference for its natural substrate. Using a combination of rational and random mutagenesis, we have engineered IDH to use isopropylmalate as a substrate. Rationally designed mutations were based on comparison of IDH to a similar enzyme, isopropylmalate dehydrogenase (IPMDH). A chimeric enzyme that replaced an active site loop-helix motif with IPMDH sequences exhibited no activity toward isopropylmalate, and site-directed mutants that replaced IDH residues with their IPMDH equivalents only showed small improvements in k(cat). Random mutants targeted the IDH active site at positions 113 (substituted with glutamate), 115, and 116 (both randomized) and were screened for activity toward isopropylmalate. Six mutants were identified that exhibited up to an 8-fold improvement in k(cat) and increased the apparent binding affinity by as much as a factor of 80. In addition to the S113E mutation, five other mutants contained substitutions at positions 115 and/or 116. Most small hydrophobic substitutions at position 116 improved activity, possibly by generating space to accommodate the isopropyl group of isopropylmalate; however, substitution with serine yielded the most improvement in k(cat). Only two substitutions were identified at position 115, which suggests a more specific role for the wild-type asparagine residue in the utilization of isopropylmalate. Since interactions between neighboring residues in this region greatly influenced the effects of each other in unexpected ways, structural solutions were best identified in combinations, as allowed by random mutagenesis.     

The Heparin-Binding Activity of Secreted Modular Calcium-Binding Protein 1 (SMOC-1) Modulates Its Cell Adhesion Properties

Secreted modular calcium-binding proteins 1 and 2 (SMOC-1 and SMOC-1) are extracellular calcium- binding proteins belonging to the BM-40 family of proteins. In this work we have identified a highly basic region in the extracellular calcium-binding (EC) domain of the SMOC-1 similar to other known glycosaminoglycan-binding motifs. Size-exclusion chromatography shows that full length SMOC-1 as well as its C-terminal EC domain alone bind heparin and heparan sulfate, but not the related chondroitin sulfate or dermatan sulfate glycosaminoglycans. Intrinsic tryptophan fluorescence measurements were used to quantify the binding of heparin to full length SMOC-1 and the EC domain alone. The calculated equilibrium dissociation constants were in the lower micromolar range. The binding site consists of two antiparallel alpha helices and mutagenesis experiments have shown that heparin-binding residues in both helices must be replaced in order to abolish heparin binding. Furthermore, we show that the SMOC-1 EC domain, like the SMOC-2 EC domain, supports the adhesion of epithelial HaCaT cells. Heparin-binding impaired mutants failed to support S1EC-mediated cell adhesion and together with the observation that S1EC in complex with soluble heparin attenuated cell adhesion we conclude that a functional and accessible S1EC heparin-binding site mediates adhesion of epithelial cells to SMOC-1.     

Tertiary structure of bacteriorhodopsin. Positions and orientations of helices A and B in the structural map determined by neutron diffraction

Positions and rotations of two helices in the tertiary structure of bacteriorhodopsin have been studied by neutron diffraction using reconstituted, hybrid purple membrane samples. Purple membrane was biosynthetically 2H-labeled at non-exchangeable hydrogen positions of leucine and tryptophan residues. Two chymotryptic fragments were purified, encompassing either the first two or the last five of the seven putative transmembrane segments identified in the amino acid sequence of bacteriorhodopsin. The 2H-labeled fragments, diluted to variable extents with the identical, unlabeled fragment, were mixed with their unlabeled counterpart; bacteriorhodopsin was then renatured and reconstituted. The crystalline purple membrane samples thus obtained contained hybrid bacteriorhodopsin molecules in which certain transmembrane segments had been selectively 2H-labeled to various degrees. Neutron diffraction powder patterns were recorded and analyzed both by calculating difference Fourier maps and by model building. The two analyses yielded consistent results. The first and second transmembrane segments in the sequence correspond to helices 1 and 7 of the three-dimensional structure, respectively. Rotational orientations of these two helices were identified using best fits to the observed diffraction intensities. The data also put restrictions on the position of the third transmembrane segment. These observations are discussed in the context of folding models for bacteriorhodopsin, the environment of the retinal Schiff base, and site-directed mutagenesis experiments.