Membrane Insertion Kinetics of a Protein Domain In Vivo

The pathway by which segments of a polytopic membrane protein are inserted into the membrane has not been resolvedin vivo. We have developed an in vivo kinetic assay to examine the insertion pathway of the polytopic protein bacterioopsin, the apoprotein of Halobacterium salinarumbacteriorhodopsin. Strains were constructed that express the bacteriorhodopsin mutants I4C:H6 and T5C:H6, which carry a unique Cys in the N-terminal extracellular domain and a polyhistidine tag at the C terminus. Translocation of the N-terminal domain was detected using a membrane-impermeant gel shift reagent to derivatize the Cys residue of nascent radiolabeled molecules. Derivatization was assessed by gel electrophoresis of the fully elongated radiolabeled population. The time required to translocate and fully derivatize the Cys residues of I4C:H6 and T5C:H6 is 46 ± 9 and 61 ± 6 s, respectively. This is significantly shorter than the elongation times of the proteins, which are 114 ± 26 and 169 ± 16 s, respectively. These results establish that translocation of the bacterioopsin N terminus and insertion of the first transmembrane segment occur co-translationally and confirm the use of the assay to monitor the kinetics of polytopic membrane protein insertion in vivo.

An essential step in the biogenesis of polytopic or multispanning membrane proteins is the insertion of polypeptide segments into the membrane. To elucidate the insertion mechanism in vivo, the cellular factors that participate in this process must be characterized, and the pathway or sequence with which the polypeptide segments are inserted must be determined.
Cellular factors have been identified that mediate protein insertion into the eukaryotic endoplasmic reticulum (1,2) and bacterial cytoplasmic membranes (3)(4)(5). These include the signal recognition particle (SRP), 1 a cytosolic ribonucleoprotein complex, and the secretory translocase, a membrane protein complex. SRP directs ribosome-bound nascent polypeptides to the secretory translocase, which in eukaryotes forms an aqueous channel into which the polypeptides are inserted (6,7). Sequence conservation of SRP and secretory translocase subunits (8,9) suggests that aspects of the insertion mechanism are universal.
Less is known about the insertion pathway of polytopic membrane proteins. In eukaryotes, a co-translational, sequential insertion pathway (10,11) is favored (12). In support of this model, the insertion of transmembrane segments of polytopic membrane proteins has been shown to be mechanistically coupled to translation (13)(14)(15) and to occur sequentially (15)(16)(17). However, other evidence contradicts this model. At least one eukaryotic polytopic membrane protein appears to insert posttranslationally (18), and several others exhibit multiple topologies that may reflect a nonsequential insertion pathway (19 -21). In bacteria, it is not known whether insertion is cotranslational and sequential or more closely resembles protein secretion, where translocation is independent from elongation (22,23). There is evidence for both post-translational (24) and co-translational (4,5,25) insertion mechanisms, but the issue remains controversial.
Studies of the insertion pathway have been problematic for several reasons. First, methods used to monitor the membrane insertion of protein segments are not time-resolved. Significant time elapses between polypeptide translation and analysis of insertion by methods such as cross-linking (14,26) or proteolysis (16). During this time, polypeptides may rearrange to yield a topology that does not reflect the state of the polypeptides during insertion. Second, the topology of most polytopic proteins is poorly defined, making it difficult to interpret insertion studies unambiguously. Third, topological reporters commonly used to study insertion, such as fusion domains (27,28) or glycosylation sequences (21), may alter protein topology or insertion. Finally, the insertion pathway determined in vitro may differ from the in vivo pathway because of differences in the concentrations of SRP or secretory translocase subunits, the membrane potential, or ionic conditions.
To examine the insertion pathway of a polytopic membrane protein in vivo, we have studied bacteriorhodopsin (BR), a light-driven proton pump of known structure (29) expressed in the sole membrane of the archaeon Halobacterium salinarum. BR contains a 248-amino acid polypeptide, bacterioopsin (BO), which forms seven transmembrane ␣-helices that surround a covalently bound retinal cofactor. The polypeptide is initially synthesized with a 13-amino acid presequence that is cleaved during biogenesis (30).
In this work, we have developed an in vivo assay of BO insertion and used it to characterize the translocation kinetics of the first extracellular domain of the protein. The assay monitors the translocation of engineered Cys residues in BO with a membrane-impermeant gel shift reagent specific for sulfhydryl groups. Comparison of the translocation time of Cys residues near the N terminus and the BO elongation time indicated that the N-terminal domain inserts co-translationally. These results establish the assay as a direct method to monitor the insertion of transmembrane segments in vivo and raise the possibility that the predominant pathway of polytopic membrane protein insertion is co-translational in all taxonomic domains.
Plasmid and Strain Construction-Mutations in bop were constructed in Escherichia coli by polymerase chain reaction. To introduce a sequence encoding six His residues at the BO C terminus, the primers TGTTGAGCGACGCTGGAAAG and GGCGACGGCGCGGCCGCGCA-CCACCACCACCACCACTGATCGCACACGCAG were used (codon changes are underlined). The last three codons of bop were removed by these changes. Constructs encoding Cys at amino acids 4 and 5 of mature BO were created with the primers GCGCGGATCCGACGTGA-AGA and CCGGACGTCCGGTGCACTGGGCCTGCGATA or CCGGAC-GTCCGCAGATCTGGGCCTGCGATA. Restriction fragments derived from the polymerase chain reaction products were cloned in pMPK62 (31) to yield bop genes encoding His-tagged BO with a single Cys near the N terminus. Recombinant H. salinarum strains expressing the BR variants were created by targeted gene replacement (31). The bop sequence of the recombinant strains was confirmed by fluorescent dye terminator cycle sequencing (ABI Prism, Foster City, CA).
Expression and Characterization of Mutant BR-To induce BR synthesis, H. salinarum was cultured with illumination (32) in 120 ml of peptone medium for 2 days (33). Cell lysates were prepared, and BR levels were determined as described (34) from the change in absorbance at 585 nm between light-and dark-adapted samples. BR levels were expressed as percentages of total cell protein determined by the BCA assay (Pierce). For spectral analysis, mutant proteins were purified (33) and scanned in 30 mM sodium phosphate, 0.025% sodium azide, pH 6.9, with a Perkin-Elmer 2 spectrophotometer.
Purification of His-tagged BO-Full-length His-tagged BO was purified from H. salinarum to Ͼ90% homogeneity by denaturing Ni 2ϩ affinity chromatography at room temperature. Cultures (120 ml) were harvested at 8,000 ϫ g for 10 min at room temperature, and the pellet was resuspended in 1.2 ml of medium salts (33). A 25-l aliquot was mixed with 1.0 ml of 60 units DNaseI/ml and 100 g phenylmethylsulfonyl fluoride/ml (lysis solution) and incubated for ϳ10 min. Samples were combined with 40 l of Ni 2ϩ -nitrilotriacetic acid beads and adjusted with buffer to a final concentration of 300 mM NaCl, 50 mM Tris-Cl, pH 7.0, 0.1% SDS in 1.4 ml. After incubation for 2 h, beads were sedimented and washed with 1 ml of 0.01% SDS, 300 mM NaCl, 50 mM Tris-Cl, pH 7.0, 5 mM imidazole. Bound protein was eluted by incubating for 0.5 h with 50 mM imidazole, 20 mM EDTA, 50 mM dithiothreitol (DTT), in Laemmli sample buffer (35). His-tagged proteins were recovered quantitatively, as determined by immunoblotting with BR-114 antibody provided by Dr. H. G. Khorana (data not shown).
Measurement of Elongation Time-To establish the elongation time of His-tagged BO, cells were radiolabeled as described (36) with the following modifications. Cells cultured to induce BR were harvested, washed, resuspended in medium salts containing 0.5% L-Ala and shaken at 40°C for 2.5 h with illumination (36). Cells were then harvested and resuspended in 1 ⁄100th volume medium salts containing 0.5% L-Ala. A 400-l concentrated cell suspension was stirred at 37°C in a glass vial exposed to Ͼ520 nm illumination. After 5 min, 160 Ci of [ 35 S]Met was added (pulse) followed 20 s later by 400 l of 10 mM nonradioactive Met in medium salts containing 0.5% L-Ala (chase). At various times, 50 l of radiolabeled cells were combined with 1.0 ml of lysis solution to halt synthesis. Full-length His-tagged BO was recovered by Ni 2ϩ affinity chromatography and electrophoresed on a 14% polyacrylamide gel (35), which was stained with Coomassie and dried. Radioactivity in BO was quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and normalized by the intensity of the Coomassie-stained bands.
The elongation rate was determined by fitting the elongation time course data with a step-wise function similar to that described (37). The maximum extent of radiolabeling was estimated by averaging points in the plateau region. The data were fit with a function containing nine steps distributed according to the spacing of Met residues in BO, with an increase in radiolabeling of 1 ⁄9th the plateau value at each step. A constant elongation rate was assumed and was the only adjustable parameter. To correct for lags due to Met equilibration, the function was fit starting at 41 s (see below and "Results"). The elongation time was calculated by dividing the length of mature His-tagged BO, 251 amino acids, by the derived elongation rate. The 13-amino acid presequence is not observed by this approach, presumably due to its rapid cleavage during biogenesis. Consequently, radiolabeling of this region is not detected and is not included in the analysis.
Measurement of the Met Equilibration Time-To determine the lag in radiolabeling due to uptake and equilibration of Met during the pulse and chase, cells were radiolabeled or pulse-chase radiolabeled with [ 35 S]Met as above. At various times, aliquots were removed and added to 40 volumes of lysis solution. Total protein in 10 l was precipitated onto filter paper with trichloroacetic acid (10% w/v). [ 35 S]Met incorporation was quantified by PhosphorImager analysis, plotted as a function of time after the pulse and fit by linear regression. The lag in [ 35 S]Met incorporation after the pulse is given by the x intercept of the fit to data from radiolabeled cells. The lag in incorporation of nonradioactive Met after the chase is given by the time between chase addition and the intersection of the curves fit to data from radiolabeled and pulse-chase radiolabeled cells.
Measurement of Elongation Time by Lag in Radiolabel Incorporation-Elongation times were also determined by a method in which the lag times preceding steady-state radiolabeling of total cell protein and a specific protein are compared (38). Cells were radiolabeled and lysed at various times after the pulse. The time course of [ 35 S]Met incorporation into total cell protein and His-tagged BO was determined as in previous sections. Linear fits of the data yielded x intercepts that correspond to the lags preceding steady-state radiolabeling of total cell protein and His-tagged BO, respectively. The difference in the x intercepts equals 1 ⁄2 the elongation time of His-tagged BO.
Measurement of the Translocation Time-To measure translocation times, cells expressing mutant BR were pulse-chase radiolabeled as above, except 2 l of 200 mM AMS in dimethyl sulfoxide were added to 400 l of the cells 1 min before radiolabeling. At various times after the pulse, 50 l of the cells were combined with 50 l of 20 mM DTT, incubated for 10 min at 37°C to complete elongation of the radiolabeled protein, and mixed with 1.0 ml of lysis solution. His-tagged protein was recovered and electrophoresed as above, except that a pH 7.5 separating gel was used, and gels were run for 21 h at 11°C at 200 V on a 20-cm gel apparatus to maximize resolution. The derivatized protein was expressed as a percentage of the total radioactivity in each lane and plotted as a function of time. These data were fit to the curve generated from fitting the elongation data, with a shift along the time axis as the only adjustable parameter. The magnitude of this shift is equal to the difference in the translocation and elongation times. The length of the polypeptide upon translocation of the Cys residue was estimated by subtracting the value of this shift from the elongation time and multiplying this difference by the elongation rate.

RESULTS
Experimental Rationale-The in vivo kinetic assay is designed to detect when a transmembrane segment inserts into the membrane relative to other events in the lifetime of a membrane protein. In the assay, insertion is monitored by the translocation of a unique extracellular Cys across the cytoplasmic membrane. Translocation is detected by derivatizing Cys with AMS, a membrane-impermeant sulfhydryl reagent that shifts protein mobility on SDS-polyacrylamide gel electrophoresis (39,40). Cells expressing the membrane protein are preincubated in AMS and then radiolabeled to monitor nascent polypeptides (Fig. 1, I). At various times after radiolabeling, derivatization is quenched with a thiol reagent, DTT (Fig. 1, II). Samples are then incubated to complete elongation of the radiolabeled population, and the full-length protein is purified (Fig. 1, III). Derivatized and underivatized protein are separated by SDS-polyacrylamide gel electrophoresis and quantified. By this approach, the fraction of the radiolabeled population in which Cys has been translocated may be determined as a function of time after translation is initiated. To assess whether insertion occurs co-or post-translationally, the translocation time course is compared with the elongation time course, measured independently by pulse-chase radiolabeling.
Expression and Characterization of Mutant Proteins-To de-velop the assay, Cys was substituted for Ile 4 or Thr 5 , located in the 8-amino acid extracellular domain at the N terminus of mature BO (41). Translocation of this domain reports on the insertion of the adjacent transmembrane ␣-helix of mature BO, which includes residues 9 -31. A hexahistidine tag was introduced at the C terminus of the wild-type and Cys mutant proteins for purification. The His-tagged Cys mutant proteins, I4C:H 6 and T5C:H 6 , were expressed in H. salinarum at significantly lower levels than wild-type BR but only slightly lower than BR:H 6 ( Table I). The proteins could be purified in a form similar to the wild-type purple membrane (33) and had normal spectral properties (Table I), indicating that they assemble normally and bind the retinal chromaphore.
Reactivity of the Mutant Proteins-To demonstrate the reactivity of the Cys residues, the mutant proteins were derivatized with AMS. A shift was observed with T5C:H 6 ( Fig. 2A) and I4C:H 6 (data not shown) but not BO:H 6 , which lacks Cys ( Fig.  2A). The reaction of AMS with T5C:H 6 in intact cells was complete in ϳ10 s and was effectively quenched by DTT (Fig.  2B). Thus, AMS is sulfhydryl-specific and reacts rapidly.
Translocation Time of the BO N Terminus-As the first step in establishing the translocation time of the N-terminal domain, AMS derivatization of T5C:H 6 and I4C:H 6 was measured at various times after radiolabeling according to the scheme in Fig. 1. Results are shown for T5C:H 6 (Fig. 3); similar results were obtained for I4C:H 6 (data not shown). A substantial fraction of radiolabeled T5C:H 6 was derivatized at the earliest time point, indicating that many of the nascent polypeptides are inserted at the start of the assay (Fig. 3A). Derivatization increased to a plateau of Ն90% (Fig. 3B), consistent with maturation of the nascent radiolabeled population.
Several observations suggest that the appearance of derivatized protein at early times is not due to derivatization of nascent BO inside the cell. First, AMS is unlikely to cross the membrane because of its two negatively charged sulfonate groups. It has been used in other systems as a membrane impermeant reagent (42)(43)(44). Second, translocation time courses similar to that in Fig. 3B were obtained when the preincubation with AMS was extended by several minutes or when the reagent was added after pulse-chase radiolabeling (data not shown). Finally, different translocation time courses are observed with Cys located in other extracellular regions of the protein, arguing that AMS derivatization reflects Cys translocation. 2 Measurement of Elongation Time-To determine whether translocation of Cys residues near the N terminus occurs co-or post-translationally, we measured the BO elongation time by labeling cells with a short pulse of [ 35 Fig.  4A; similar results were obtained with I4C:H 6 and BO:H 6 (data not shown). A plot of the data, normalized for the total number of Met residues in the protein (Fig. 4B, open circles), was fit with a curve that accounts for the distribution of Met in BO  a BR expression levels from 2-day-old cultures were determined by light-dark spectroscopy of crude cell lysates as described (34). Values reported are the average of two experiments.
b Absorption maxima of the purified lattice form of the proteins were obtained in 30 mM sodium phosphate, pH 6.9. Dark-adapted, incubated for Ͼ8 h at room temperature in the dark; light-adapted, illuminated for 5 min with Ͼ520 nm light. Aliquots were removed at various times and quenched with an equal volume of 20 mM DTT in medium salts. Full-length T5C:H 6 was purified by Ni 2ϩ affinity chromatography and electrophoresed. The percentage of derivatized, radiolabeled T5C:H 6 was determined by PhosphorImager analysis. A monoexponetial fit of the data yielded a time of ϳ10 s for 99% completion. (Fig. 4B, dashed line). Prior to fitting, the data were corrected for lags due to pulse and chase equilibration, which require 19 Ϯ 3 and 22 Ϯ 2 s, respectively (Fig. 4C). From the curve, the average elongation rates and corresponding elongation times for I4C:H 6 and T5C:H 6 were found to be within ϳ30% of BO:H 6 (Table II). Although the source of this variation has not been identified, the effect on the calculation of translocated chain length is minimal (see below). The elongation values for I4C:H 6 and T5C:H 6 are in excellent agreement with those obtained by an independent method (Table II,  The BO elongation rates (Table II) are lower than measured rates in eukaryotes (3-6 amino acids/s (45)) and bacteria (12-20 amino acids/s (46)). To determine whether the BO elongation rate was atypical, the elongation rates of several prominent radiolabeled proteins in cell lysates (M r ϭ ϳ24 -57 kDa) were measured by the pulse-chase radiolabeling method used for BO. Elongation rates from 1.3 to 5.5 amino acids/s were obtained (data not shown), indicating that translation is relatively slow in H. salinarum and that BO translation is not unusual.
Comparison of the Translocation and Elongation Times-Translocation and elongation data were compared to determine when the T5C:H 6 N-terminal domain translocates relative to synthesis of the full-length protein (Fig. 4B, closed circles). Strikingly, Cys translocation occurs significantly before elongation; similar results were obtained for I4C:H 6 . Thus, translocation of the N-terminal extracellular domain and insertion of the first transmembrane ␣-helix of BO occur co-translationally in H. salinarum.
The translocation and elongation data were used to estimate the length of the polypeptide upon translocation of the Cys residue. Because both data sets reflect BO maturation, they were fit by similar curves, except that the fit of the translocation data was displaced along the time axis. The fits of the translocation and elongation data for T5C:H 6 are shown in Fig.  4B; similar analysis was applied to I4C:H 6 . By this analysis, translocation of the Cys residue in T5C:H 6 occurs more than 70 s before elongation is complete (Table II, columns 1 and 2). Multiplying the time required for translocation by the average elongation rate (Table II), the length of the T5C:H 6 polypeptide upon translocation of the Cys residue was calculated to be 92 Ϯ 12 amino acids, assuming that the polypeptide is translated at a uniform rate. A similar result was obtained for I4C:H 6 (Table  II). Notably, even though the elongation rates differ, the calculated translocated chain length agrees. Correcting for the ϳ10 s required for complete derivatization (Fig. 2B), the calculated chain length upon translocation of the Cys residue in T5C:H 6 is 77 Ϯ 12 amino acids. DISCUSSION We have developed a kinetic assay to monitor the membrane insertion of polytopic proteins and used it to demonstrate that the first transmembrane segment of BO inserts into the H. salinarum membrane co-translationally. To our knowledge, the . At various times, aliquots were removed and precipitated with trichloroacetic acid (10% w/v). Radioactivity in trichloroacetic acid-precipitable total protein was measured at each time point in duplicate and fit by linear regression. The time required for equilibration of the pulse and chase was calculated as described under "Experimental Procedures." insertion kinetics of a polytopic membrane protein have not been established in vivo in any other organism. The assay was feasible because the H. salinarum membrane is accessible to the external environment, AMS reacts with translocated Cys very rapidly, and BO translation is sufficiently slow to resolve translocation events. Although H. salinarum is ideal for this approach, it may be possible to apply the method to other prokaryotes in which the cytoplasmic membrane is accessible and translation is slow. Examination of other Cys substitutions in BO with this assay will be invaluable for determining the complete insertion pathway of this protein.
Our results suggest that the insertion pathway of an archaeal polytopic membrane protein is similar to the co-translational pathway proposed for eukaryotic polytopic proteins (10,11). In this pathway, ribosome-bound nascent polypeptides must reach a length of ϳ70 amino acids to traverse the membrane (47,48). A similar length of 77 Ϯ 12 amino acids is synthesized in T5C:H 6 before the Cys residue is translocated, suggesting that ribosomes synthesizing BO are in close proximity to the membrane. The similarity between the membrane insertion of archaeal and eukaryotic proteins may reflect the conservation of "information pathways" such as transcription and translation in archaea and eukaryotes (49). These results, together with evidence of co-translational insertion in bacteria (4,5,25), raise the possibility that the predominant mode of polytopic membrane protein insertion is co-translational in all three taxonomic domains.
The finding that translocation of the BO N terminus is coupled to translation has important implications for understanding BR folding. Because the N-terminal transmembrane segment is the first region of the mature protein to achieve a transmembrane configuration, it may interact with other regions of BO to promote their folding within the membrane. Recent evidence that the first two helices of BO facilitate the folding of the remaining helices supports this idea (50). Cotranslational insertion of the first transmembrane segment may nucleate folding of the remaining transmembrane segments and must be considered in studies of BO folding in vitro.
Co-translational translocation of the BO N terminus is consistent with both a translocase-dependent mechanism, in which protein factors within the lipid bilayer mediate insertion (10,11), and a spontaneous mechanism, in which no such factors are involved (51,52). Although our results do not distinguish between these possibilities, we propose that BO insertion is mediated by SRP and the secretory translocase, as has been demonstrated for eukaryotic and bacterial membrane proteins. This is reasonable, given that H. salinarum SRP is implicated in BO biogenesis (53) and that homologs of secretory translocase subunits exist in several archaea (9), including H. salinarum. 3 Direct studies are required to test whether these cellular factors mediate BO insertion. a Elongation time and rate were determined as described in the legend to Fig. 4. b Elongation time was determined by the difference in the lag times of [ 35 S]Met incorporation into total trichloroacetic acid-precipitable protein and full-length BO.