Technical Comments

Response to Comment on “Principles of ER cotranslational translocation revealed by proximity-specific ribosome profiling”

See allHide authors and affiliations

Science  12 Jun 2015:
Vol. 348, Issue 6240, pp. 1217
DOI: 10.1126/science.aaa8299


Reid and Nicchitta propose that most cellular translation is carried out by a noncycling pool of endoplasmic reticulum (ER)–associated ribosomes. However, proximity-specific ribosome profiling data place an upper bound of about 7 to 16% on the fraction of cytosolic protein translation carried out by ribosomes accessible to ER-tethered biotin ligases. Moreover, yeast pulse-labeling experiments argue against there being a static population of ER-associated ribosomes.

Reid and Nicchitta have raised a fundamental question in biology: Where in the cell does translation occur? In what represents a pronounced departure from the current textbook thinking (1), they argue that “the ER is a primary site of general protein synthesis” (2) and that ribosomes stay stably associated with the endoplasmic reticulum (ER) through many rounds of translation. As they point out, if the ER were the primary site of translation of both secreted and cytosolic proteins, this would complicate the interpretation of our ribosome exchange experiments. However, analysis of our data reveals that the large majority of cytosolic proteins are translated by a pool of ribosomes that are inaccessible to ER-tethered biotin ligases (BirA), whereas a rapidly exchanging pool of ER-associated ribosomes strongly favors translation of secreted proteins.

At the heart of this issue is the interpretation of our proximity-specific ribosome profiling enrichment data. The relative proportion of translation of nonsecretory proteins that is carried out by ER-associated ribosomes (i.e., ribosomes that are accessible to ER-tethered biotin ligases) can be calculated as β from the following equation relating secretory footprint representation in our input and pulldown librariesθS,pulldown = θS,input/[θS,input + β(1 – θS,input)]where θS is the fraction of footprints from ribosomes translating secretory proteins (S) in the input (θS,input) or streptavidin pulldown (θS,pulldown) libraries, and β is the fraction of footprints from nonsecretory proteins in streptavidin pulldown libraries [x: (x ∉ S)] (3, 4). Importantly, β provides an upper bound on translation of nonsecretory proteins at the ER because these footprints can be from bona fide translation at the ER, nonspecific labeling of cytosolic ribosomes, or background binding. Indeed, it is likely that a low level of background from the much larger pool of translation of cytosolic proteins contributes substantially to β.

We calculated β for experiments labeling ribosomes with the three different forms of ER-tethered BirA (BirA-Ssh1, Sec63-BirA, and BirA-Ubc6). This analysis revealed that at most 7 to 16% of nonsecretory protein translation occurred on ribosomes that were accessible to ER-tethered BirA (Table 1). Importantly, this is inconsistent with the proposal by Reid and Nicchitta (5) that BirA-Ubc6 biotinylates all ribosomes at the ER surface (i.e., those translating both cytosolic and secreted protein), whereas BirA-Ssh1 only biotinylates ribosomes translating secretory proteins after associating with the translocon, because the β’s for the two BirA fusions are comparable. An interesting open question is whether there is a “peri-ER” pool of ribosomes that remain in the vicinity of the ER but are not accessible to any of the ER-tethered BirA proteins.

Table 1 Proportion of secretory footprints in data sets labeled for 7 min with cycloheximide.

View this table:

Reid and Nicchitta also present an analysis of the codon-specific enrichment of secretory proteins. We disagree with this analysis on two grounds. The first is a conceptual one. They argue that it was the observation that enrichment in BirA-Ssh1 labeling occurs only after the emergence of the signal sequence that led us “to conclude that ribosomes are cotranslationally targeted from the cytosol to the ER after the signal sequence is translated.” This conflates the question of which mRNAs are cotranslationally targeted with at what point during translation targeting occurs. In fact, in our paper, we addressed the question of cotranslational translocation [figure 3 in (3)] before the introduction of position-specific analyses [figure 4 in (3)], which was motivated by a distinct set of questions. Moreover, in the Jan et al. Research Article (3), we argue that our data are best explained by a model in which, after an initial round of targeting, mRNAs of secreted protein remain tethered to the ER by downstream translating ribosomes [figure 5C in (3)]. Thus, a substantial fraction of the time, ribosomes will initiate translation on messages for secretory proteins that are tethered to the ER.

The second disagreement is a technical one. Despite claiming to “reproduce the authors’ primary observation,” the metagene enrichments presented by Reid and Nicchitta both qualitatively (position dependence) and quantitatively (magnitude of enrichment) fail to recapitulate our results [compare Reid and Nicchitta’s figure 1 (5) with our Fig. 1 and with figure 4B in (3)]. It is not clear to us how they arrived at their plots, because insufficient details were provided and they purport to analyze data sets (“7-min biotin pulse without cycloheximide” for BirA-Ssh1/BirA-Ubc6) that do not exist. In any case, it is clear that they have not correctly replicated our analysis. We have reproduced our results with three independent analyses implemented by two individuals, and all biotin ligase fusions show position-specific enrichment that maximizes after the emergence of the signal sequence/anchor (Fig. 1, code available on request).

Fig. 1 Metagene analyses.

(A) Metagene plot of log2 BirA-mVenus-Ssh1 enrichment per codon (mean ± SD) as a function of ribosome position relative to the first codon of the first Phobius-predicted hydrophobic element for secretome proteins (excluding tail-anchored proteins). The heat map below represents single-gene enrichments sorted by the number of codons before the first hydrophobic element. Cells were treated with 100 μg/ml cycloheximide and incubated with biotin for 2 min. (B) As in (A) for BirA-mVenus-Ubc6 enrichment. (C) As in (A) for Sec63-mVenus-BirA enrichment.

Fig. 2 Kinetic analysis.

(A) Bar graph showing the percentage of footprints aligning to secretome proteins in streptavidin pulldown libraries from the indicated BirA-fusion protein after a 7-min incubation with biotin. The dashed line shows the average percentage of secretome footprints in the input libraries. (B) Plotted are the percentages of footprints in pulldown libraries that aligned to cotranslationally targeted secretome genes (circles) or all other genes (triangles). Horizontal dashed lines indicate the percentage of footprints mapping to secretome (black) and nonsecretome (green) genes in whole-cell ribosome profiling libraries. The x axis indicates translation time after biotin addition.

Reid and Nicchitta also argue that ribosomes are stably associated with a static rough ER. We observed a rapid loss of the enrichment for secreted proteins when biotinylation was allowed to proceed with ongoing translation (3). Reid and Nicchitta argue that this is not due to exchange of ribosomes from the ER after translation termination and new initiation. Rather, based on their model that noncycling ER-associated ribosomes translate mostly cytosolic proteins, they propose that there are different kinetics of labeling from ER-localized BirA of ribosomes translating secreted proteins and those translating cytosolic ones due to “a modest kinetic advantage for ribosomes in close proximity to translocon-associated BirA-Sec63.” However, this interpretation is ruled out by the persistent enrichment of secretome proteins in the biotinylated pool through extended labeling times when translation elongation, and thus ribosome exchange, is inhibited with cycloheximide (Fig. 2A) [see also figure 2A in (3)]. Thus, it is the act of translation termination and subsequent initiation on a new message that allows ribosomes labeled on the ER surface to gain efficient access to the pool of cytosolic messages.

Additionally, the kinetic analysis presented by Reid and Nicchitta in figure 2 in (5) is incorrect. The correct plot of the composition of footprints in pulldown libraries during our pulse-labeling studies reveals a robust enrichment for secretory protein that rapidly decays when labeling is allowed to proceed in the presence of ongoing translation (Fig. 2B). Importantly, although there is marked enrichment after 1 min of labeling, this enrichment is substantially lower than when labeling is carried out with cycloheximide (Fig. 2B). Thus, even after 1 min (roughly the time it takes for a single round of translation), there is already considerable ribosome exchange that cannot be accounted for by differences in BirA labeling kinetics.

Finally, Reid and Nicchitta inaccurately characterize our conclusions as being “consistent with the established signal hypothesis/signal recognition particle (SRP) pathway model.” Our study was not meant to be a test, let alone a validation of the standard SRP model. Indeed, some of our findings clearly go beyond the canonical SRP model, including the observations that SRP-independent proteins are cotranslationally targeted to the ER, that the duration of translation after the emergence of a signal sequence/anchor is a major determinant of cotranslational translocation, and that mRNAs can be targeted to the ER even in the absence of SRP (3). Also, as discussed above, we argue that after the initial recruitment of a secretory protein message to the ER, it can be retained there through tethering by multiple ribosomes [figure 5C in (3)]. In summary, our data demonstrate that, regardless of the initial targeting mechanism, the majority of secretory proteins are translated at the ER, whereas the large majority of cytosolic proteins are not.


  1. We define S as the set of secretory proteins enriched above a receiver operator characteristic threshold that maximized accuracy for Sec63-BirA enrichment [figure S5 in (3)].
View Abstract

Navigate This Article