The Permeability of the Endoplasmic Reticulum Is Dynamically Coupled to Protein Synthesis*

Proteins synthesized by the rough endoplasmic reticulum (RER) co-translationally cross the membrane through the pore of a ribosome-bound translocon (RBT) complex. Although this pore is also permeable to small molecules, it is generally thought that barriers to their permeation prevent the cyclical process of protein translation from affecting the permeability of the RER. We tested this hypothesis by culturing Chinese hamster ovary-S cells with inhibitors of protein translation that affect the occupancy of RBTs by nascent proteins and then permeabilizing the plasma membrane and measuring the permeability of the RER to a small molecule, 4-methyl-umbelliferyl-α-d-glucopyranoside (4-MαG). The premature or normal release of nascent proteins by puromycin or pactamycin, respectively, increased the permeability of the RER to 4-MαG by 20–30%. In contrast, inhibition of elongation and the release of nascent proteins by cycloheximide did not increase the permeability, but it prevented the increase in permeability by pactamycin. We conclude that the permeability of the RER is coupled to protein translation by a simple gating mechanism whereby a nascent protein blocks the pore of a RBT during translation, but after release of the nascent protein the pore is permeable to small molecules as long as an empty ribosome remains bound to the translocon.

enough to be permeated by many other small molecules when a ribosome-bound translocon (RBT) is translationally inactive and empty, i.e. the pore is not occupied by a nascent protein.
We previously demonstrated (2) that the permeability of RBTs to a small, polar molecule (4-methyl-umbelliferyl-␣-D-glucopyranoside (4-M␣G)) was increased after puromycin, a tRNA analog, prematurely terminated translation and released nascent protein chains from the PCC (path a, Fig 1). The increased permeability to 4-M␣G was consistent with a previous report that RBTs incorporated into planar bilayers and opened by puromycin are permeable to ions (3), and our results were also supported by a recent report that puromycin can release calcium from the RER (4). The permeation of empty RBTs is especially significant in the context of recent studies by Nicchitta and colleagues (5,6), who reported that approximately two-thirds of the 60 S subunits remain bound to translocons after the normal completion of protein translation, thereby constituting a large pool of persistent, translationally inactive RBT complexes. Although permeation of these empty RBTs could have very important consequences for RER-dependent signaling and the maintenance of essential gradients across the RER membrane, this pathway has received relatively little attention.
The permeation of translocons by small molecules has also been investigated by Johnson and colleagues (7,8), who concluded that a permeability barrier is maintained at the luminal end of the translocon by the binding of BiP, a luminal chaperone protein, and that the junction between the 60 S subunit and the translocon is tightly sealed. Their conclusions have led to the broadly held view that RBTs are tightly sealed at all times, preventing the process of protein translation from influencing the permeability of the RER to small molecules. We propose, however, that the notion that the permeability of the RER is not affected by permeation of RBTs might be limited to experimental conditions in which translationally inactive, empty ribosomes have been stripped from ER microsomes by a high-salt wash. This experimental detail is important because it limits the detection of permeation to only two states of the translocon, the translationally active 80 S-bound translocon and the ribosome-free translocon (blocked and closed states, respectively, in Fig. 1). A novel feature of our current study is that we have used experimental conditions in which permeation of a third state of the translocon, translationally inactive 60 S-bound RBTs, can also be detected.
In this article we examine two models that describe the relationship between the translational activity of RBTs and the permeability of the RER to a small molecule, 4-M␣G. The first model is a "static permeability" model in which RBTs are never permeable to small molecules other than nascent proteins, and the permeability of the RER is therefore independent of translational activity. The second model is a "dynamic permeability" model in which the pore of an RBT is "gated" closed when it is occupied by a nascent protein, but the pore of translationally inactive, 60 S-bound translocons is open and permeable to small molecules. This simple physical mechanism for coupling the permeability of the RER to protein translation predicts a dynamic relationship in which increasing the level of protein translation and the occupancy of RBTs by nascent proteins should decrease the permeability of the RER, whereas inhibiting the initiation of protein translation and decreasing the occupancy of RBTs by nascent proteins should increase the permeability of the RER.
The appropriateness of the static versus dynamic permeability models can be tested by determining the extent to which the permeability of the RER is dependent on the level of protein translation and the occupancy of RBTs by nascent proteins. The static permeability model predicts complete independence, whereas the dynamic permeability model predicts that the permeability of the RER is inversely related to the level of occupancy by nascent proteins during protein translation. The goal of the current study was to test these predictions by measuring the permeability of the RER to 4-M␣G while manipulating protein translation with drugs that inhibit protein translation by different mechanisms of action. Puromycin (PUR) resembles aminoacyl-tRNA, and it is a substrate for the peptidyltransferase of the 60S subunit, accepting a nascent peptide and causing the premature release of incomplete nascent proteins (9). Pactamycin (PAC) inhibits the initiation of protein translation by preventing formation of the 43S initiation complex, but it permits the normal elongation and release of full-length proteins whose synthesis was initiated prior to addition of the drug (10,11). Cycloheximide blocks elongation and prevents the normal completion and release of nascent proteins from RBTs (12). Cycloheximide stabilizes polyribosomes and the association of 80S ribosomes with translocons, whereas PUR and PAC elicit rapid breakdown of polyribosomes, but leave the majority of translationally inactive 60S subunits bound to the RER (5). We report that the permeability of the RER was sensitive to the manipulation of protein translation and changes in the occupancy of RBTs by nascent proteins, and we conclude that the permeability of the RER is dynamically coupled to protein translation.

EXPERIMENTAL PROCEDURES
Materials-4-M␣G, cycloheximide, and puromycin-HCl were from Calbiochem. Pactamycin was a generous gift from The Upjohn Co. All other reagents were from Sigma.
Cell Culture-CHO-S cells (Invitrogen), a cell line derived from Chinese hamster ovary K1 cells and selected for growth in suspension, was used for most experiments. A cell line derived from HEK-293 cells for growth in suspension (HEK-293-F) was also obtained from Invitrogen. HepG2 and BHK-21 cells were obtained from the ATCC (Manassas, VA). CHO-K1 cells were a gift from Ed Levitan (University of Pittsburgh). CHO-S and HEK-293-F cells were grown in serum-free medium (CHO-SFII and 293SFII medium, respectively, from Invitrogen). CHO-K1 cells were grown in Ham's F12 medium containing 5% serum, and BHK-21 and HepG2 cells were grown in Dulbecco's modified Eagle's medium containing 10% serum. All cells were cultured at 37°C, 8% CO 2 .
4M␣G Assay-The 4-M␣G assay was performed as described by Heritage and Wonderlin (2). Briefly, CHO-S cells and HEK-293-F cells were typically grown to a density of 0.5-1.0 ϫ 10 6 cells/ml in T-75 flasks or in spinner flasks. For a 32-well assay (eight conditions in quadruplicate), 20 ml of cells was pelleted and resuspended in 20 ml of K-G buffer (140 mM potassium glutamate, 2.5 mM MgCl 2 , 10 mM HEPES, pH 7.25) and gently permeabilized by N 2 cavitation (2 min, 80 p.s.i.). This procedure was modified for adherent cells. BHK-21 and CHO-K1 cells were grown in T-75 flasks, harvested by scraping, and then homogenized in K-G buffer by N 2 cavitation. HepG2 cells were grown to confluence in 48-well plates, and the medium was replaced with K-G buffer containing 0.08% digitonin (100 l/well) for 5 min to permeabilize the cells and then diluted 1:5 with K-G buffer. Nunc 48-well plates were loaded with 0.5 ml of sample/well, the plate with solutions was prewarmed to 35°C, and 4-M␣G (20 M from a 20 mM stock in methanol) was added immediately before placing the plate into a CytoFluor 4000 plate reader. The fluorescence was measured for 30 min at 2-min intervals with 10 s mixing before each measurement. The center wavelengths/bandwidths of the excitation and emission filters were 360/40 and 460/40, respectively.
Data Analysis-In CHO-S cells, 4-M␣G is activated when it enters the lumen of the RER and it is hydrolyzed by ␣-glucosidase II. The slope of the fluorescence versus time curve, S (t) , is proportional to the rate of entry and activation of the dye at time t. The linear and exponential contributions to S (t) were fitted by Equation 1, where S 0 is the initial slope (⌬F min Ϫ1 ) and k is an exponential rate constant (min Ϫ1 ). Best-fit estimates of S 0 and k for each well were obtained using the Solver nonlinear curve-fitting routine in Excel (Microsoft). All experimental conditions were assayed in parallel Ϯ detergent (0.05% Igepal, Fig. 2A) so that the permeability to 4-M␣G under a specific experimental condition could be expressed as a dimensionless fractional permeability (P), where S 0,max is the rate of activation after the RER membrane was permeabilized by detergent and the access of 4-M␣G to the ␣-glucosidase II was maximal. Statistical analysis was performed using JMPin (SAS Institute). Averages are plotted Ϯ standard error of the mean. Calcium Measurements-CHO-S cells were loaded with 10 M Fluo-3-AM (30 min, 23°C) and then washed twice in Hanks' balanced salt solution. Fluorescence emission at 510 nm was measured with excitation at 485 nm in a CytoFluor 4000 plate reader. Adherent BHK-21 cells grown in 10% Dulbecco's modified Eagle's medium (8% CO 2 , 37°C) were released by trypsinization and gentle trituration and then loaded with 10 M Mag-Fura-2-AM (60 min, 37°C) in Dulbecco's modified Eagle's medium minus serum. The cells were washed twice in Hanks' balanced salt solution and loaded into a stirred cuvette in a SPEX Fluorolog spectrometer. The ratio of fluorescence measured at 510 nm with excitation alternating between 340 and 380 nm was calculated at each time point without further calibration.

RESULTS
Rationale for Testing the Permeability Models-We have studied the permeation of RBTs in CHO-S cells by 4-methylumbelliferyl-␣-D-glucopyranoside (4-M␣G), a small, membrane-impermeant probe; it's entry into the RER can be detected when it is hydrolyzed to a fluorescent product by luminal ␣-glucosidase II. We previously reported (2) that the activation of 4-M␣G in intact CHO-S cells was negligible, but the selective rupture of the plasma membrane by N 2 cavitation enabled 4-M␣G to access the RER, and it revealed a substantial basal permeability of the RER to 4-M␣G (P basal ). This permeability was further increased to a level termed P PUR when nascent chains were released by the addition of PUR (Fig. 2). The static permeability model predicts that P basal should be insensitive to changes in the occupancy of RBTs by nascent proteins produced by concomitant changes in protein translation. This model was tested using a two-step procedure. In the first step, cells were pretreated in culture with PUR, PAC, or CHX to produce changes in the occupancy of RBTs. In the second step, these cells were permeabilized, split into two samples, and P was determined in each sample either without PUR (P basal ) or with PUR (P PUR ) added to the sample, respectively. The occupancy of translationally active RBTs by nascent proteins after the pretreatment period was estimated by calculating the PUR-dependent increase in P, where ⌬P PUR is proportional to the number of blocked, translationally active RBTs that could be opened by PUR added during the assay. If pretreatments that decreased the occupancy of RBTs by nascent proteins (as inferred from a significant decrease in ⌬P PUR ) also significantly increased P basal , the static model should be rejected. Given the importance of P basal and ⌬P PUR in these tests, we first examined in greater detail three features of P basal and ⌬P PUR in this system: (1) the similarity of P basal and ⌬P PUR measured in CHO-S cells to other cell lines, (2) the specificity of ⌬P PUR as a measure of the occupancy of translationally active RBTs by nascent chains, and (3) the contribution of RBTmediated permeation to P basal .
Similarities in Permeability (P) among Different Cell Lines-We examined the activation of 4-M␣G in the absence and presence of PUR in CHO-S, CHO-K1, HEK-293-F, BHK-21, and HepG2 cells. We observed in all of these cell lines a substantial P basal and a similar 30 -40% increase in P when puromycin was added to the assay (Fig. 2B). The similarity between CHO-S and HepG2 cells is especially notable, because Seiser and Nicchitta (5) reported that approximately twothirds of the ribosomes remain bound to the RER membrane in HepG2 cells following either the premature termination of translation by PUR or the normal completion of translation with pactamycin. We conclude from these similarities that the mechanisms that regulate the permeation of RBTs are probably also similar among these cell lines and that the high P basal and the ⌬P PUR are not unusual features of CHO-S cells. These results support our use of CHO-S cells in the present study as a model system for investigating the permeability of the RER.
⌬P PUR Is a Measure of the Pool of Translationally Active RBTs-We previously concluded (2) that ⌬P PUR was produced by the release of nascent chains from translationally active RBTs on the basis of the generally recognized mechanism of action of puromycin as a tRNA analog (9), the similar sensitivity to high salt of ⌬P PUR and the binding of ribosomes to ER membranes (13), and the absence of ⌬P PUR in the presence of detergent, which demonstrated that ⌬P PUR required the entry of 4-M␣G into a membrane-bound compartment. In the current study we wanted to use the size of ⌬P PUR to track changes in the size of the pool of translationally active RBTs (i.e. those RBTs occupied by nascent proteins that could be released by PUR) in response to treatments with inhibitors of protein synthesis. The puromycin reaction catalyzed by the peptidyltransferase is essential for the release of nascent proteins, and we performed the following additional experiments to provide a more rigorous test of the specific dependence of ⌬P PUR on the puromycin reaction.
Anisomycin is an inhibitor of the peptidyltransferase and the puromycin reaction (14). When added simultaneously with PUR, 50 M anisomycin completely inhibited ⌬P PUR , but it did not inhibit ⌬P PUR when it was added 10 min after PUR, sufficient time for the irreversible release of nascent chains by PUR (Fig. 3A). The failure of anisomycin to significantly inhibit ⌬P PUR when added after PUR demonstrated that the ability of anisomycin to inhibit ⌬P PUR when applied with PUR did not result from nonspecific effects, such as blocking the pore of empty RBTs. We conclude from the complete inhibition of ⌬P PUR when anisomycin was applied with PUR that the production of ⌬P PUR required an anisomycin-sensitive peptidyltransferase reaction.
The puromycin reaction is maximum at pH 8.5-9.5, with a pK a of 7.2-7.6 and a steep decrease in rate at neutral pH and below (15,16). We observed that ⌬P PUR was inhibited when the buffer was acidified from pH 7.2 to pH 6.6 before the addition of PUR, but acidification to pH 6.6 15 min after the addition of PUR had no effect (Fig. 3B). The temporal dependence of the inhibition by low pH leads us to conclude, again, that the inhibitory effect was on the puromycin reaction and not the permeability pathway.
We conclude that the production of ⌬P PUR is entirely dependent on the puromycin reaction catalyzed by the peptidyltransferase center, and it is therefore a valid measure of the release of nascent proteins from translationally active RBTs. The measurement of ⌬P PUR provides an alternative approach for Lines are best fits (see Equation 1) to the fluorescence measured without detergent (squares) or with detergent (0.05% Igepal, circles). The addition of 100 M PUR (black) increased the slope (S 0 ) relative to control (gray) in the absence but not the presence of detergent. B, average P basal (gray) and P PUR (black) values for five cell lines. The greater variability of the averages in CHO-K1 and BHK21 cells probably resulted from the more difficult and less consistent permeabilization of these adherent cells, which increased the inter-assay variability in the base-line permeability. In contrast, the ratio P PUR /P basal was much more consistent across assays. Therefore, we tested the hypothesis that puromycin increased P by examining the ratio P PUR /P basal , which was significantly greater than a value of 1 for all cell lines (p Ͻ 0.05, multiple t tests with Bonferroni adjustment).
determining the size of the pool of translationally active RBTs that is much simpler than other methods (e.g. radiolabeling nascent proteins). It is also insensitive to translationally active ribosomes in the cytosol, because only the release of nascent proteins from translationally active ribosomes that are bound to the RER can increase the activation of 4-M␣G. Although ⌬P PUR provides only a relative measure of the translational activity of RBTs, that was sufficient for our current study in which we wanted to use ⌬P PUR to track changes in the size of the pool of translationally active RBTs in response to inhibitors of protein synthesis.
The Contribution of RBTs to P basal -We have observed over a 2-year period that P basal can slowly change in CHO-S cells over the course of several weeks or months, with P basal ranging from 0.2 to 0.7. This variability led us to speculate that the entry of 4-M␣G via pathways other than RBTs might contribute a labile background component to P basal and cause us to underestimate the importance of changes in P basal mediated specifically by permeation of RBTs. In the absence of drugs that can specifically inhibit the permeation of RBTs, we have estimated the component of P basal contributed by the permeation of RBTs, relative to the background permeability contributed by other, unidentified pathways, by examining the sensitivity of P basal to high salt, which has been reported to release up to 85% of the translationally inactive ribosomes from the RER membrane in rat hepatocytes (path c, Fig. 1) (13). Although high salt might also inhibit RBT-independent pathways for the entry of 4-M␣G, we have demonstrated previously (2) that both the concentration of salt producing half-maximal block and the slope of the salt concentration-inhibition curve were identical for ⌬P PUR and the salt-sensitive component of P basal . We surmised from the remarkably similar sensitivities to high salt that both salt-sensitive components were produced by the entry of 4-M␣G through a common pathway. Cells were split into two samples, with one sample broken open in high-salt (300 mM K) K-G buffer and the other sample broken open in normal (140 mM K) K-G buffer. A plot of P basal measured in high-salt versus normal buffer is shown in Fig. 3C. The dotted line with unity slope indicates the relationship expected if P basal was not affected by high salt, and it is clear that P basal measured in high salt was depressed relative to P basal measured in normal salt, consistent with our previous report that P basal was sensitive to high salt (2). Most importantly, a regression line fitted to these data had a slope of 0.94 (Ϯ0.10), insignificantly different from a value of 1. We conclude from the parallel downward shift of the relationship that P basal was decreased on average by Ϸ0.2 in high-salt relative to normal-salt buffers regardless of the size of P basal in normal salt. The data in Fig. 3C also indicate that the apparent lower limit of P basal is 0.2, from which we can conclude that in some samples of cells 4-M␣G enters the RER exclusively through empty RBTs. These data indicate that the size of the salt-sensitive permeability to 4-M␣G was surprisingly constant compared with the high variability of the saltinsensitive component.
Effect of Puromycin-The primary test of the static permeability model was to determine whether pretreating cells with PUR prior to permeabilization could reversibly increase P basal . CHO-S cells were pretreated with 200 M PUR for 15 min to terminate translation and release nascent proteins from RBTs, after which half of the cells were permeabilized and assayed. Pretreatment with PUR significantly increased P basal while substantially reducing ⌬P PUR (Fig. 4A), as expected if all of the available nascent proteins were released from translationally active RBTs during the pretreatment period. If the increase in P basal resulted from the release of nascent proteins and the opening of RBTs, it should be reversed by restarting translation and increasing the occupancy of RBTs by nascent proteins. The remaining control and PUR-treated cells were pelleted, washed, and transferred to control medium for 30 min, at which time P basal was not significantly different from the control level (Fig. 4A), demonstrating that restarting protein synthesis could reverse the increase in P basal .
Effect of Pactamycin-To ensure that the increase in P basal produced by PUR was not an artifact of the premature release of truncated proteins by PUR, we also examined pactamycin, which should release full-length proteins. A 1-h pretreatment with PAC significantly increased P basal and eliminated ⌬P PUR When added with PUR, anisomycin significantly decreased ⌬P PUR (p Ͻ 0.01, t test, asterisk) relative to the control. B, inhibition of ⌬P PUR by acidification. A sample of cells was permeabilized in K-G buffer acidified with glacial acetic acid to pH 6.6, and 100 M PUR was then added to half of the sample (left pair of bars, n ϭ 6). A second sample of cells was permeabilized in normal K-G buffer (pH 7.2), and 100 M PUR was added to half of the sample. After 15 min, concentrated glacial acetic acid was added to lower the pH to 6.6. (right pair of bars, n ϭ 2). Gray and black bars represent P basal and P PUR , respectively. P PUR was inhibited only when acidification occurred prior to the addition of PUR (p Ͻ 0.05, t test, asterisk) C, plot of P basal measured in paired samples of CHO-S cells broken open in either 140 K-G buffer (abscissa) or 300 K-G buffer (ordinate). The dotted line has a slope of unity, and the solid line is a line fitted by linear regression (slope ϭ 0.94, intercept ϭ Ϫ0.16, n ϭ 26). The slope of the fitted line was not significantly different from the slope of the dotted line. These data were collected over a 2-year period. (Fig. 4B), as observed with PUR pretreatment. Similar, significant increases in P basal were observed following pretreatments with PAC for 5, 30, 60, or 360 min, indicating that the increase in P basal was both rapid and sustained (data not shown).
The increase in P basal was eliminated when control and PACtreated cells were permeabilized in a buffer in which the potassium glutamate concentration was increased to 300 mM (Fig.  4B), which supports a role for empty RBTs because translationally inactive ribosomes can be stripped from the RER by a high-salt buffer (13). Also, unlike PUR, the acute application of PAC to permeabilized control cells did not increase P basal (data not shown), which is consistent with the ability of PUR to release nascent proteins directly versus the PAC requirement that synthesis be completed and nascent proteins be released prior to permeabilization.
The similar increase in P basal produced by pretreating intact cells with either PAC or PUR can be accounted for by the conversion of blocked RBTs to open RBTs. The increase in P basal was balanced entirely by the loss of ⌬P PUR , resulting in no net increase in P PUR (Fig. 4, A and B), which was expected if the increase in P basal depended solely on the conversion of blocked RBTs to open RBTs along path a in Fig. 1. Also, the increase in P basal produced by both PUR and PAC was about 70% of the ⌬P PUR measured in control cells (Fig. 4, A and B, where ⌬P PUR reflects the size of the translationally active pool in control cells), which was remarkably consistent with the previous report of a persistent binding of Ϸ65% of 60 S subunits after the release of nascent proteins (5).
Effect of Cycloheximide-In contrast to PUR and PAC, a 1-h pretreatment with 200 M cycloheximide (CHX) did not increase P basal (Fig. 4C). A similar lack of effect was observed with 5-or 360-min pretreatments (data not shown), indicating that the different effects of PAC and CHX were not limited to a specific point in time. The failure of CHX to increase P basal demonstrated that the increased P basal produced by PUR or PAC was not a nonspecific consequence of inhibiting protein translation (e.g. clearing of proteins from the lumen of the RER). The fact that PAC, but not PUR, requires elongation for the release of nascent chains led us to predict that CHX would prevent the increase in P basal by PAC, but CHX should have much less of an effect on the increase in P basal by PUR. Cotreatment with CHX completely prevented PAC from increasing P basal (Fig. 4C), whereas CHX decreased ⌬P PUR by only 20 -30% (data not shown).
PUR-dependent Release of Calcium in Intact Cells-We also examined changes in the permeability of the ER to Ca 2ϩ to determine whether changes in the permeability to 4-M␣G could be generalized to Ca 2ϩ and as a control to ensure that changes in the permeability of the ER could be observed in intact cells in which no cytosolic constituents that might regulate the permeability of RBTs could be lost, in contrast to the 4-M␣G assay that requires permeabilization of the plasma membrane for the entry of 4-M␣G. Intact CHO-S cells were loaded with the Ca 2ϩ indicator Fluo-3 and treated in parallel with PUR and thapsigargin (TG), a selective inhibitor of the Ca-ATPase in the ER (17), with the response to TG providing a positive control for a Ca 2ϩ signal produced by the release of Ca 2ϩ from the ER. Thapsigargin (100 nM) and PUR produced similar, triphasic changes in [Ca 2ϩ ] (Fig. 5, A and B), both beginning with a transient increase. The increase in [Ca 2ϩ ] elicited by PUR was not a nonspecific effect of inhibiting protein synthesis, because it was observed in cells in which protein synthesis was already inhibited by pretreatment with CHX (Fig. 5A). We also used Mag-Fura-2 to monitor more specifically changes in the permeability to Ca 2ϩ . BHK-21 cells were loaded with Mag-Fura-2 under conditions in which the [Ca 2ϩ ] in the lumen of the ER could be preferentially monitored (18). Although we could not detect changes in luminal [Ca 2ϩ ] with the addition of PUR to intact BHK-21 cells (data not shown), PUR significantly increased the rate of release of Ca 2ϩ from the ER in intact cells when re-uptake was inhibited by TG (Fig. 5C), evidence that the permeability to Ca 2ϩ was increased. Our observation that PUR increased the permeability of the RER to calcium in intact cells is consistent with a recent report that PUR depleted calcium in the RER of dialyzed or permeabilized cells (4), and our observations in intact cells demonstrated that the changes in the permeability to 4-M␣G produced by PUR did not result from the loss of potential regulatory factors following permeabilization of the plasma membrane. DISCUSSION We conclude from the experiments reported above that the permeability of the RER can be dynamically coupled to the level of protein translation by changes in the number of empty (i.e. open) versus blocked RBTs. Dynamic coupling could be produced by a simple mechanism requiring only the following features: (1) the pore of the PCC and the exit tunnel of the 60 S subunit are blocked by a nascent protein when the RBT is translationally active; (2) following the normal completion of translation and the release of a nascent protein, some empty ribosomes remain persistently bound to translocons; (3) the pore of empty RBTs is permeable to small molecules; and (4) the relative distribution of blocked versus empty translocons is influenced by the level of protein translation. The apparent high barrier to the permeation of translocons by small molecules, as described by Johnson and colleagues (Ref. 7; also reviewed in Ref. 8), is probably limited to experimental conditions in which permeable, translationally inactive ribosomes have been stripped from translocons by high salt, leaving only impermeable, translationally active RBTs and closed, ribosome-free translocons. This conclusion is supported by our current and previous observation (2) that P basal was significantly lower when measured in a high-salt buffer, presumably a result of the dissociation of ribosomes from translationally inactive RBTs.
At a macroscopic level, the functional consequences of the dynamic coupling between P basal and protein translation depend on the dynamic range over which P basal can vary, and we expect that this range will be affected by the properties of different permeant molecules. For 4-M␣G, the release of all of the nascent chains by PUR or PAC increased P basal by Ϸ30% (we have not yet examined the level to which P basal can be decreased by increasing translational activity and converting empty RBTs to blocked RBTs). However, the presence of both salt-sensitive and salt-insensitive components of P basal measured with 4-M␣G provided evidence that it could enter the RER through one or more channels or carriers in addition to empty RBTs, which was not surprising given the small size and simple structure of 4-M␣G. The increased background permeability resulting from the permeation of more than one pathway by 4-M␣G reduced the apparent size of the increase in P basal that was produced by the release of nascent proteins. For larger molecules that can enter the RER by permeating empty RBTs but not via smaller channels or carriers, the dynamic range could be significantly larger.
The dynamic coupling between P basal and protein translation appears to occur in addition to a substantial contribution of empty RBTs to the steady-state permeability of the RER. We measured the salt-sensitive component of P basal , which includes, but is not necessarily limited to, the entry of 4-M␣G through empty RBTs. From these measurements we estimated that permeation of salt-sensitive, empty RBTs in the absence of puromycin consistently contributed a fractional permeability to 4-M␣G of 0.2 regardless of the size of P basal (Fig. 3C). Given an average size of ⌬P PUR of 0.10 (2), we estimate that the ratio of empty to occupied RBTs might be as high as Ϸ2:1. This ratio is considerably higher than expected based on ultrastructural studies in which most ribosomes appear to be part of translationally active polysomes (e.g. Fig. 174 in Ref. 20). However, our results are more consistent with a study by Adelman et al. (13) in which ϳ40% of the ribosomes could be released from rat liver microsomes with salt alone, and an additional 40 -45% could be released with the combination of high salt and puromycin. These authors concluded that the ribosomes released by high salt in the absence of puromycin were mostly translationally inactive, empty ribosomes, and their results would support a ratio of empty to occupied RBTs of Ϸ1:1. Our results are also consistent with a high permeability to small solutes observed in rough microsomes prepared from rat liver without treatment with high salt (19). Resolution of this issue will require parallel measurements of permeability and ultrastructure in the same cells.
A high resting permeability to small molecules mediated by empty RBTs also raises the question as to how the RER might maintain essential gradients across its membrane or different environments (e.g. oxidizing versus reducing) in the lumen versus the cytosol. Although the Ca-ATPase in the ER membrane could maintain a calcium gradient in the presence of a leak pathway, it is less clear which transporters might maintain gradients for other molecules of interest. Of course, any speculation as to the potential effect of empty RBTs on the permeability of the RER to small molecules must be tempered by the fact that thus far we have only demonstrated permeation by 4-M␣G and calcium, and further study will be required to determine the repertoire of small molecules that can permeate empty RBTs.
The macroscopic changes in P basal described above might be especially important during global changes in protein translation, such as when the initiation of protein synthesis is inhibited in response to cellular stress (21). In this context, the increase in P basal produced by PAC provides a new clue into the mechanism of stress-induced apoptosis. Many cytosolic and ER stressors inhibit the initiation of protein translation by activating the phosphorylation of the translation initiation factor eIF-2␣ (21), an effect mimicked by PAC. If inhibiting the initiation of translation increases the permeability of the RER, as we observed with PAC, then RER-dependent homeostatic processes might be disrupted, thereby promoting apoptosis. The FIG. 5. PUR increased the permeability to calcium. A, CHO-S cells were loaded with Fluo-3, and PUR or TG (100 nM) was added at time zero. A single representative experiment is shown with cells pretreated for 30 min with 50 M CHX to inhibit protein synthesis prior to the addition of PUR. B, averaged responses in five experiments performed as described in A and plotted with standard error bars. For TG, the latency to the decline from the peak to the nadir was highly variable. C, decay of ER luminal Ca 2ϩ in BHK-21 cells loaded with Mag-Fura-2. 100 nM TG alone or TG and 200 M PUR were added at time zero. The decay of luminal [Ca 2ϩ ] was fitted with a single exponential function and plotted with the fitted line on a semi-log plot. The inset shows a significant decrease (p Ͻ 0.05, t test, n ϭ 6) in the average time constant for the decay with PUR (black) and without PUR (gray). ability of cycloheximide to stabilize nascent chains within RBTs and prevent PAC from increasing P basal might explain why CHX can also prevent stress-induced apoptosis under conditions in which protein synthesis is already inhibited, such as during ischemia/reperfusion (22); and the opposing actions of CHX and PUR on P basal might account for the paradoxical, pro-apoptotic effects of PUR in situations where CHX is protective (23,24). From a methodological perspective, experimental manipulations that stress cells and inhibit the initiation of protein synthesis (e.g. removing extracellular calcium) could have large effects on the permeability of the RER.
Dynamic fluctuations in P basal might also occur at a microscopic level as protein synthesis within local regions of the RER undergoes its normal cyclical course. For example, the permeation of empty RBTs by calcium or other small molecules following the release of a completed protein could produce a diffusible signal coupled to the end of translation, perhaps playing a role in stimulating the disassembly and reassembly of a translationally active RBT with successive cycles of protein translation. Molecules entering the lumen through an empty RBT might also signal its availability for the retrograde export of misfolded proteins through the pore of the RBT (25). At a microscopic level, the opening of RBTs might produce nearly quantal changes in P basal compared with the macroscopic changes in P basal averaged over larger dimensions. For example, the opening of an RBT could produce a rapid and large change in the local concentrations of molecules within the lumen of the RER, given the very small volume of the lumen and the potentially large flux of molecules through an empty RBT.
In conclusion, the permeability of the RER to small molecules should be viewed as a dynamic property influenced by protein translation. We propose that dynamic coupling could occur in any cells in which empty ribosomes remain bound to translocons following the completion of translation. Recognition of the dynamic nature of this relationship provides new insight into the generation of novel signals and the potential loss of homeostatic regulation under pathophysiological conditions.