Protein Translocation through and Insertion into Membranes

Co-translational Trafficking at the ER Membrane: In mammals, secretory and most membrane proteins are synthesized by ribosomes bound to sites at the membrane of the endoplasmic reticulum (ER) termed translocons where the nascent proteins are translocated across or integrated into the ER membrane co-translationally. To monitor directly the environment and interactions of the nascent protein chains (i.e., to “see” what the nascent chain sees), we incorporate fluorescent, photoreactive, or other probes into the nascent chains in vitro at specific sites using modified aminoacyl-tRNAs (aa-tRNAs) that have a probe covalently attached to the aa side chain. The incorporation of non-natural amino acids into protein in vitro using modified aa-tRNAs is an experimental approach that I originated more than 35 years ago to examine protein secretion, and it has since been used by many labs for a variety of purposes.

Our unique fluorescence approaches have provided novel perspectives on various aspects of protein trafficking and have yielded many insights. Among other things, we have shown that: nascent secretory proteins are translocated through the ER membrane via an aqueous, gated pore in the translocon; the inner diameter of the pore is 40-60 Å when bound to the ribosome during translocation and is 9-15 Å after the ribosome is released; and the ribosome, not the translocon, first recognizes that a nascent chain is a membrane protein and then initiates the conversion of the translocon operational mode from translocation to integration. We also discovered that the permeability barrier of the ER membrane is maintained during protein trafficking by the highly coordinated binding of the ribosome to the cytoplasmic side of the translocon and of BiP to the lumenal side of the membrane to effect closure of the cytosolic or lumenal end of the aqueous pore of the translocon, respectively. We are currently investigating the following questions, among others: How is the permeability barrier of the ER membrane maintained during the co-translational integration of a multi-spanning membrane protein at the translocon? Is the pore closed alternately by lumenal BiP and the cytoplasmic ribosome as transmembrane sequences or segments (TMSs) with opposite orientations are integrated? At what point does every other TMS in a multi-spanning membrane protein rotate by 180° before moving laterally into the bilayer? Does a TMS invert as a rigid α-helix or in an unfolded form? When are TMSs released from the translocon and immersed in the bulk lipids of the ER membrane?

Post-translational Trafficking at Mitochondrial Membranes: We have recently established another unique experimental system to examine the conformation of a core protein in a functional translocon in an intact membrane. By replacing, one at a time, 54 of 222 residues of Tim23 with Cys and then using seven independent fluorescence and other techniques to analyze the environment and interactions of each of these sites after the Tim23 derivative had been imported into active mitochondria and assembled into functional TIM23 translocon complexes, we have characterized the structure and conformational changes of Tim23 at a high resolution that is unprecedented. One segment of the Tim23 protein (TMS2) was shown to fold into an α-helix in the inner mitochondrial membrane with one side facing an aqueous environment and the other side in a nonpolar milieu. Chemical crosslinking showed that the aqueous-facing surface of TMS2 also contacted a protein being imported into the mitochondrion. These combined data revealed that the aqueous pore through which proteins are transported through the TIM23 complex is formed in part by a specific helical surface of Tim23p TMS2. In addition, TMS2 was shown to undergo major conformational changes when the electrochemical potential across the inner mitochondrial membrane was dissipated, and these changes were reversible. Also, the quaternary structure of the TIM23 complex was dramatically changed during both protein import into the matrix and the dissipation of the membrane potential, and these changes were also reversible. These structural changes in active, intact inner mitochondrial membranes show that the translocon is anything but a passive hole in the membrane through which proteins move.

Nascent Protein Folding

To determine how the ribosome distinguishes between nascent secretory and membrane proteins (see above), we created an experimental approach that directly detects and monitors the folding of ribosome-bound nascent chains using FRET. This technique revealed that a nascent membrane protein, but not a nascent secretory protein, folds far inside the ribosomal nascent chain exit tunnel near the peptidyltransferase center. The ribosome-induced folding of a TMS in the nascent chain into an α-helix (or nearly so) in the tunnel is then communicated via ribosomal and translocon proteins to the other side of the membrane to regulate BiP gating of the lumenal end of the aqueous translocon pore. We have since found that the ribosome induces the folding of each of the TMSs in a multi-spanning membrane protein far inside the nascent chain tunnel, and that this nascent chain folding triggers structural changes on both sides of the ER membrane that maintain the ER membrane permeability barrier during integration. The success of this novel FRET approach demonstrates that many previously-inaccessible structural, mechanistic, and regulatory questions in protein trafficking can now be addressed experimentally, as well as issues in protein folding, chaperone function, protein processing and assembly, and quality control, among others. This approach will also reveal where and when nascent membrane proteins assemble and fold into native structures during integration at the translocon.

Misfolded Protein Retro-translocation into the Cytosol for Degradation

After 10+ years there is no consensus on the molecules and mechanisms that mediate ER-associated degradation (ERAD), the movement of misfolded or misassembled secretory and membrane proteins from the ER lumen through the ER membrane to the cytosol [“retro-translocation” (RT) or “dislocation”] for degradation because of ambiguities in the various in vivo and in vitro experimental approaches used to date. We therefore designed and established a new in vitro approach that directly monitors ERAD substrate RT through the ER membrane continuously and in real time. Fluorescent-labeled ERAD substrates are biochemically encapsulated in mammalian ER microsomes along with the desired lumenal proteins and small molecules, and the reconstituted microsomes are then incubated in a solution containing the desired cytosolic proteins and small molecules with antibodies that quench the fluorescence of any cytosolically exposed dye. Substrate RT is initiated and synchronized by raising the sample temperature to 30°C, and substrate passage through the ER membrane is detected by the quenching of its dye in the cytosol. Since even subtle changes in continuous, real-time RT quenching rates can be quantified as a function of the cytosolic, lumenal, or membrane components in these well-defined samples, this approach allows the molecules and mechanisms that mediate the RT of different substrates to be directly identified. We are currently characterizing the cytosolic and lumenal requirements for the RT of two different polypeptides, one an Ig κ light chain and the other the plant toxin ricin. We are also determining whether ERAD substrates are folded or unfolded during RT, and if the latter, which end, if either, goes first.

Bacterial Toxins

Some bacterial protein toxins function by binding to the surface of mammalian cells, inserting into the bilayer, and creating holes in the membrane that lead to cell death. Perfringolysin O (PFO) is secreted by Clostridium perfringens, the pathogenic bacteria that cause gas gangrene. PFO binds to cholesterol-containing membranes and oligomerizes to form huge pores with diameters of ~300 Å. In collaboration with Dr. Rodney Tweten at the University of Oklahoma Health Sciences Center, the PFO gene was site-specifically mutated to position a single cysteine residue within the protein at various locations, and a fluorescent dye was covalently attached to the Cys in each purified mutant protein to yield (in most cases) a functional cytolytic protein. By using this approach and multiple independent fluorescence techniques, we have discovered that PFO insertion into the bilayer creates an extended amphipathic β-sheet that forms the pore-membrane interface, and that each PFO monomer contributes two membrane-spanning β-hairpins to this large β-barrel. The spontaneous and cooperative insertion of PFO into the membrane involves major conformational changes, including a unique conversion of six short α-helices into four anti-parallel β-strands, and a rearrangement of domains that moves the domain that spans the membrane more than 100 Å. Kinetics experiments reveal that pore formation requires an obligatory sequence of interactions between the membrane and two spatially separated domains of PFO, as well as inter-domain communication.

We have also examined the mechanism by which the PFO molecules prevent premature association into oligomers and thereby regulate the timing of pore formation. The toxin is secreted as a soluble monomeric protein, and it is stable as a soluble monomer until it encounters and binds to a membrane that contains sufficient cholesterol. A cholesterol-dependent conformational change that extends more than 70 Å then exposes the edges of a β-sheet in each monomer that allows them to associate with each other and thereby form the oligomer in the proper alignment to create the β-barrel. We also discovered that pore formation required the stacking of aromatic side chains in these two β-strands, thereby explaining for the first time how β-hairpins in adjacent monomers of β-barrel forming oligomeric proteins are aligned at the proper angle to insert and form the β-barrel. We are now determining at what point during pore formation the 6 α-helices unfold to form 2 β-hairpins.

Blood Coagulation

Blood clot formation is both accomplished and regulated by proteins that circulate in the blood in an inactive form until being cleaved and activated by a highly-specific proteolytic enzyme. Seven of these enzymatic reactions occur on a membrane surface, and in each case, a physiologically significant reaction rate is obtained only in the presence of a non-enzymatic protein cofactor (Fig. 1) that interacts with a specific enzyme and stimulates its activity to either promote or prevent clot formation. Our FRET measurements have shown that the elongated proteins in these complexes bind to the membrane at one end and project approximately perpendicularly from the surface, with the enzyme active sites positioned more than 70 Å above the surface (the distances determined by crystallography are in excellent agreement with our FRET-determined distances measured years earlier). One of our primary goals was to determine how the cofactors dictate both the activity and the specificity of the enzymes. To this end, we discovered that cofactor binding to its cognate enzyme alters the height and/or orientation of the active site relative to the membrane surface in all but one case. We later showed directly that the cofactor requirement for enzyme activity could be eliminated by lowering the enzyme's active site. (Fig. 2) Thus, cofactors regulate blood coagulation, at least in part, by positioning the enzyme active sites at the proper height and orientation above the phospholipid surface to cleave their membrane-bound substrates. This previously-unrecognized topographical mechanism for regulating the activity of a membrane-bound enzyme may also be used in other enzyme systems.

Protein Biosynthesis

Our use of functional aminoacyl-tRNAs (aa-tRNAs) carrying a fluorescent, photoreactive, or chemically reactive probe revealed several important aspects of ribosome and elongation factor structure and function. For example, the relative orientation of the two tRNAs bound to the A and P sites of the ribosome was determined using FRET, and the negatively-charged tRNAs were found to be surprisingly close to each other, aligned approximately side-by-side with their midsections separated by less than 15 Å. This measurement also demonstrated that the magnitude of the conformational change associated with the EF-G-dependent step in protein biosynthesis is considerable, with the midsection of a tRNA moving about 26 Å during its translocation from the A site to the P site. This early FRET study accurately measured the tRNA-tRNA distance more than 18 years before the crystal structure of a ribosome·tRNA complex was determined (the FRET and crystal structure distances differ by less than 3 Å). Similarly, our early investigations of membrane-bound ribosomes using fluorescence and photocrosslinking revealed that nascent chains move through a long aqueous tunnel in the large ribosomal subunit. The affinity of aa-tRNA for EF-Tu•GTP, determined at equilibrium using fluorescence spectroscopy, is much higher than is necessary to obtain aa-tRNA•EF-Tu•GTP ternary complex formation in the cell. This result indicates that the tight binding is necessary to keep the ternary complex intact during the initial stages of codon-anticodon recognition at the ribosome prior to GTP hydrolysis. We also showed that different aa-tRNAs had a variety of conformations at their "elbows" when free in solution, but that the aa-tRNA conformations were the same or very similar after binding to EF-Tu•GTP. This observation strongly indicates that one role of elongation factor EF-Tu (and presumably EF-1α) in protein biosynthesis is to minimize the diversity of aa-tRNA conformations that must be evaluated, presumably with uniform and optimal accuracy, at a single decoding site on the ribosome during the aa-tRNA selection process.

Finally, as discussed above, we discovered that the ribosomal tunnel is not simply an inert teflon tube designed to direct nascent chain emergence from the ribosome away from the sites where tRNA-mRNA and factor-ribosome interactions occur. Instead, nascent chain interactions with ribosomal components exposed on the surface of the tunnel regulate  translocon pore closure during membrane protein integration with membrane-bound ribosomes, as well as translation itself for cytoplasmic ribosomes synthesizing SecM. In each case, the regulatory effects were initiated by nascent binding to and folding in the ribosomal tunnel.