Regulation of Translation of ATF4 mRNA: A Focus on Translation Initiation Factors and RNA-Binding Proteins
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe review « Regulation of translation of ATF4 mRNA: a focus on translation initiation factors and RNA-binding proteins » by P. Adjibade and R. Mazroui describe several aspect of the translational regulation of the Transcription factor ATF 4 in response to various strees.
In the context of this reviewing job, I have not verified the very long bibliography cited and I mainly focused on the form of the review.
General comments :
The review is generally well written and appear quite complete to my understanding. However, some parts are quite dense and hard to read. A general effort to format the text in more readable paragraphs would improve it.
In particular the part 4.1.1 on PUS7 requires some reformating to simplify the understanding. This could be done by separating the paragraph adressing the indirect effect of PUS7 on MCTS1 and the paragraph adressing the direct effect of PUS7 on ATF4 mRNA. It is to be noted that this whole paragraph on PUS 7 seems to be supported by only one reference [144].
The iconography could generally be improved and homogenized. In particular a figure setting together the different actors of the translation initiation that are described in the text could help to better understand the impact on TC formation or availability. This could be a combination of figure 1 and 2.
The description of the regulation of ATF4 by uORF could be introduced by citing examples of other mRNA controled by uORF. An interesting recent review on uORF could be cited. https://doi-org.insb.bib.cnrs.fr/10.1016/j.csbj.2024.10.042.
Another aspect that is notadress in the review is the specificity of the translational regulation of ATF4. Is there any or other examples of mRNAs that are translated commonly with ATF4 ?
In chapter « 4 Non canonical eIFs », the description of eIF2D, DENDR and MCTS1 is confusing. Nothing is said on the fact that MCTS1 and DENDR form an heterodimeric complex. This could help to introduce the concept tha eIF2D is composed of functional domains homolog to MCTS1 (N-term) and homolog to DENDR (C-term).
I do not understand why the « RNA modifying enzyme chapter (4.1) » is hierarchically under the « 4. Non canonical eIFs » shouldn’t it be a (5) then 5.1 PUS7, 5,2 demethylases and 6. conclusions.
Specific comments :
line 71 : « TO locate the first codon » . This imply that the 43S PIC has the « will » to find a start codon. This could be changed to « AND locate the first codon »
line 111 : « and instead found associated ». I would correct to « and instead are found associated ».
line 284 : « disrupts decamer formation ». It could be interesting to precise « disrupt eIF2B decamer formation » as it may not be obvious that eIF2B is a decamer in particular when it is indicated (line 220) that eIF2B is composed of 5 subunits.
line 534-538 : What is MCTH1 ? It is not a known gene. Is it MCTS1 ?
line 574 : « This increases » => « This increase »
line 584 : It could be interesting to state rightaway that ALKBH5 is a m6A RNA demethylase.
line 600 : Citing the orginal publication that described FTO as a demethylase would be useful.
line 653 : « a faction » => « a fraction »
line 703 : « that inhibits global inhibition of translation initiation via eIF2 phosphorylation ». shouldn’t it read « that inhibits global translation initiation via eIF2 phosphorylation »
Author Response
1- The part 4.1.1 on PUS7 requires some reformating to simplify the understanding
Response: We have separated the section 4.1.1 into multiple paragraphs.
2-The iconography could generally be improved and homogenized. In particular a figure setting together the different actors of the translation initiation that are described in the text could help to better understand the impact on TC formation or availability. This could be a combination of figure 1 and 2.
Response: We have combined figure 1 with 2
3-The description of the regulation of ATF4 by uORF could be introduced by citing examples of other mRNA controlled by uORF.
Response: We have included examples of ATF4-related mRNAs in the introduction section.
4-In chapter « 4 Non canonical eIFs », the description of eIF2D, DENDR and MCTS1 is confusing. Nothing is said on the fact that MCTS1 and DENDR form an heterodimeric complex. This could help to introduce the concept that eIF2D is composed of functional domains homolog to MCTS1 (N-term) and homolog to DENDR (C-term).
Response: This is now clarified in the corresponding new section 3.6
5-I do not understand why the « RNA modifying enzyme chapter (4.1) » is hierarchically under the « 4. Non canonical eIFs » shouldn’t it be a (5) then 5.1 PUS7, 5,2 demethylases and 6. conclusions.
Response: We have corrected the numbering of our sections.
Specific comments :
1-line 71 : « TO locate the first codon » . This imply that the 43S PIC has the « will » to find a start codon. This could be changed to « AND locate the first codon »
Response: This is corrected
2-line 111 : « and instead found associated ». I would correct to « and instead are found associated ».
Response: This is corrected
3-line 284 : « disrupts decamer formation ». It could be interesting to precise « disrupt eIF2B decamer formation » as it may not be obvious that eIF2B is a decamer in particular when it is indicated (line 220) that eIF2B is composed of 5 subunits.
Response: This is corrected
4-line 534-538 : What is MCTH1 ? It is not a known gene. Is it MCTS1
Response: This is corrected
5-line 574 : « This increases » => « This increase »
Response: This is corrected
6-line 584 : It could be interesting to state rightaway that ALKBH5 is a m6A RNA demethylase.
Response: This is corrected
7-line 600 : Citing the orginal publication that described FTO as a demethylase would be useful.
Response: This is corrected
8-line 653 : « a faction » => « a fraction »
Response: This is corrected
9-line 703 : « that inhibits global inhibition of translation initiation via eIF2 phosphorylation ». shouldn’t it read « that inhibits global translation initiation via eIF2 phosphorylation »
Response: This is corrected
We thank the reviewer for their constructive comments, corrections and suggestions.
Reviewer 2 Report
Comments and Suggestions for AuthorsThis review article provides a comprehensive and insightful overview of the regulatory mechanisms governing ATF4 mRNA translation. By synthesizing 177 references and 11 well-designed figures, it effectively guides the reader through canonical uORF-mediated delayed reinitiation, the involvement of diverse translation initiation factors, and the role of RNA modifications . The clear distinction between the canonical integrated stress response (c-ISR) and the recently discovered split-ISR (s-ISR) is particularly noteworthy and provides a timely update to the field. While the manuscript is strong in its descriptive coverage of activation factors, several major areas require further development to provide a truly integrative and mechanistically complete perspective. Therefore, a major revision is recommended.
1. The authors should incorporate the negative feedback mechanisms that terminate ATF4 translation into their discussion. While the manuscript covers factors that activate ATF4 translation, it lacks a detailed analysis of signal attenuation. The GADD34 (PPP1R15A)-mediated feedback loop, wherein GADD34 recruits PP1c phosphatase to dephosphorylate eIF2a and restore eIF2B GEF activity, is a fundamental component of the ISR that directly dictates the duration of the ATF4 translational window. Incorporating this feedback mechanism would provide mechanistic completeness, especially given the detailed discussion already provided for eIF2B regulation .
2. The biological significance of precise ATF4 translational control merits further emphasis in the Introduction. ATF4 orchestrates a transcriptional program with dualistic outcomes: transient activation typically promotes cell survival , whereas prolonged or chronic activation drives pro-apoptotic programs through downstream effectors such as CHOP. This dual nature underscores why the precision of ATF4 mRNA translation is critical for cellular fate. A brief articulation of this concept would strengthen the rationale for the detailed mechanistic discussions that follow.
3. The discussion regarding pharmacological modulators targeting the ATF4 translational axis should be expanded. While ISRIB is discussed as a tool compound that antagonizes the effects of peIF2a on eIF2B, other classes of agents also directly influence this pathway. These include PERK inhibitors (e.g., GSK2606414) that block upstream eIF2a phosphorylation, and compounds that prolong ISR activation by inhibiting eIF2a dephosphorylation, such as Salubrinal or Sephin1. Discussing these agents would enhance the translational relevance of the review for researchers in drug discovery.
4. The manuscript would be significantly strengthened by the inclusion of an integrative figure (Figure 12) illustrating the complete regulatory network of ATF4 mRNA translation. While the current 11 figures effectively illustrate individual components, a unified conceptual framework is currently missing. This proposed figure should depict the pathway from various stress sensors to translational output, distinguishing between acute, chronic, and intrinsic stress conditions, while incorporating the aforementioned feedback loops and pharmacological intervention points.
5. Regarding the discussion of RNA-binding proteins (RBPs), the authors should consider including TBL2 (transducin b-like 2). It has been demonstrated that TBL2 binds ATF4 mRNA via its WD40 domain and regulates its translation during ER stress (Tsukumo et al., J Cell Biochem. 2016;117:500-509). Specifically, TBL2 associates with PERK at the ER membrane, providing a unique spatial regulatory dimension to ATF4 translation. Given that the manuscript title explicitly focuses on RNA-binding proteins , the inclusion of TBL2 alongside DDX3 would provide more comprehensive coverage of this category.
6. The authors could consider incorporating a brief discussion of IRES-dependent translation as an alternative pathway for ATF4 expression. Recent research has demonstrated that DKC1-mediated pseudouridylation of 28S rRNA recruits hnRNP A1, which acts as an IRES trans-acting factor to sustain ATF4 translation under conditions of severe stress where cap-dependent initiation is significantly compromised (Gupta et al., Sci Adv. 2025;11:eadv9401). This mechanism would elegantly complement the authors' current discussion regarding PUS7-mediated pseudouridylation and provide a more integrated perspective on how diverse RNA modification pathways influence translation.
7. Although the manuscript cites the recent study by Smirnova et al. (Cell Rep. 2024;43:113976) concerning m6A validation , the discussion regarding stem-loop-induced ribosome queuing as a fine-tuning mechanism remains underdeveloped. The authors should consider expanding on how the stem-loop structure within the uORF2/ATF4 overlap region regulates ribosomal movement. Furthermore, the manuscript should mention the finding that the ATF4 start codon is subject to significant leaky scanning, which may lead to the production of N-terminally truncated ATF4 isoforms with distinct functions.
8. Improving internal cross-referencing between related sections would enhance readability. Several concepts are introduced in early sections before being fully elaborated later. For example, eIF3 is mentioned in the context of DDX3-CBC complexes before its comprehensive discussion in Section 3.5 , and ALKBH5 is introduced as an eIF3d target before its detailed mechanistic discussion . Adding brief cross-references would help readers navigate the complex regulatory network more effectively.
9. The visual presentation of the figures requires improvement for clarity. Currently, illustrations such as Figure 7 incorporate a range of heterogeneous elements—molecular ligands, pharmacological agents, kinases, and clinical phenomena like carcinogenesis—without clear visual differentiation . Adopting a standardized visual convention (e.g., consistent shapes or color-coding for different categories) and establishing a formal legend for arrow types (e.g., solid for induction, blunt-ended for inhibition) would significantly improve figure interpretability.
10. The reference to unpublished data on lines 178-179 should be avoided in a peer-reviewed review article. Finally, the use of question marks in Figure 6B and Figure 8 to indicate uncertain mechanisms appears less authoritative; these could be replaced with dashed lines or alternative visual conventions indicating proposed interactions.
Author Response
1-The authors should incorporate the negative feedback mechanisms that terminate ATF4 translation into their discussion.
Response: We have now incorporated this feedback mechanism in the section 3.2:
2-The biological significance of precise ATF4 translational control merits further emphasis in the Introduction.
Response: We have now further stressed the biological significance of ATF4 translation in the introduction section.
3-The discussion regarding pharmacological modulators targeting the ATF4 translational axis should be expanded.
Response: We have inserted a new paragraph discussing regulation of peIF2a with PERKi and Salubrinal in section 3.2
4-The manuscript would be significantly strengthened by the inclusion of an integrative figure (Figure 12)
Response: We have now included an integrative figure 11
5-The authors could consider incorporating a brief discussion of IRES-dependent translation as an alternative pathway for ATF4 expression.
Response: A paragraph describing the role of DKC1-hnRNPA1 in ATF4 translation is included in section 4.1.1.
6-Although the manuscript cites the recent study by Smirnova et al. (Cell Rep. 2024;43:113976) concerning m6A validation, the discussion regarding stem-loop-induced ribosome queuing as a fine-tuning mechanism remains underdeveloped. The authors should consider expanding on how the stem-loop structure within the uORF2/ATF4 overlap region regulates ribosomal movement. Furthermore, the manuscript should mention the finding that the ATF4 start codon is subject to significant leaky scanning, which may lead to the production of N-terminally truncated ATF4 isoforms with distinct functions.
Response: This review focus on proteins regulating ATF4 translation. Discussing recent advance of cis-acting regulatory elements of ATF4 mRNA involved in regulating its translation is highly relevant deserving by itself a full review to be well covered.
7- Improving internal cross-referencing between related sections would enhance readability.
Response: We have now improved internal cross-referencing between sections.
8-The visual presentation of the figures requires improvement for clarity. Currently, illustrations such as Figure 7 incorporate a range of heterogeneous elements—molecular ligands, pharmacological agents, kinases, and clinical phenomena like carcinogenesis—without clear visual differentiation . Adopting a standardized visual convention (e.g., consistent shapes or color-coding for different categories) and establishing a formal legend for arrow types (e.g., solid for induction, blunt-ended for inhibition) would significantly improve figure interpretability.
Response: We have now improved the visual presentation of our figures par adopting a standardized visual convention
9-The reference to unpublished data on lines 178-179 should be avoided in a peer-reviewed review article. Finally, the use of question marks in Figure 6B and Figure 8 to indicate uncertain mechanisms appears less authoritative; these could be replaced with dashed lines or alternative visual conventions indicating proposed interactions.
Response: We removed the reference to unpublished data and replaced question marks by dashed lines.
We thank the editor and reviewers for their constructive comments, corrections and suggestions.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have made substantial efforts to address most of the concerns. The expanded Introduction (p.1-2, lines 41-50) now clearly articulates the dual nature of ATF4 signaling, establishing a strong rationale for the mechanistic discussions that follow. The incorporation of GADD34-mediated negative feedback (p.8, lines 300-305), pharmacological modulators including PERK inhibitors and Salubrinal (p.9-10, lines 352-377), and DKC1-mediated IRES-dependent translation (p.15, lines 621-628) significantly enhances the review's comprehensiveness. The new integrative Figure 11 (p.19) effectively summarizes the complete regulatory network during both canonical and split-ISR. The improved internal cross-referencing (p.6, line 213; p.13, lines 535-536), removal of unpublished data references (p.5, line 186), merging of original Figures 1 and 2 into a single Figure 1, correction of typographical errors in Figure 7 legend, and replacement of question marks with dashed lines (Figures 7 and 9) appropriately address presentation concerns. The authors' decision to decline expanding on stem-loop-induced ribosome queuing (Comment #6), focusing on trans-acting factors, is acceptable.
However, none of the figure legends include formal explanations of the symbols used throughout the manuscript, such as the meaning of arrow types (solid lines, dashed lines, blunt-ended inhibition lines), color coding (red/green), or shape conventions. The authors should add symbol legends to clarify these visual elements for readers.
Author Response
1-However, none of the figure legends include formal explanations of the symbols used throughout the manuscript, such as the meaning of arrow types (solid lines, dashed lines, blunt-ended inhibition lines), color coding (red/green), or shape conventions. The authors should add symbol legends to clarify these visual elements for readers.
Responses: We added symboles legends to all figures clarifying visual elements.
