PREreview of Epigenetic deprogramming driven by disruption of CIZ1-RNA nuclear assemblies in early-stage breast cancers

This review reflects comments and contributions by Gliday Yuka, Femi Arogundade, Arpita Ghosh, Morufu Olalekan Raimi, John Laver, Akanksha, Joseph Biggane, Rodolfo Aramayo and Vaishnavi Nagesh. Review synthesized by Anna Oliveras.

In this manuscript, the authors investigate how CIZ1 influences early-stage breast cancers by characterizing its role in forming RNA-dependent supramolecular assemblies. They report that overexpression of the CIZ1 C-terminal anchor domain (AD) turns into significant disruption RNA assemblies, which can lead to chromatin deprotection and changes in gene expression that may predispose early-stage tumors to epigenetic instability. The research integrates human cell lines, mouse models, and bioinformatics, highlighting the significance of RNA-protein interactions in cancer. The findings suggest CIZ1 is crucial in breast cancer development and offer insights into the mechanisms of epigenetic regulation.

Positive aspects of the study: 

• The study's validity is strengthened by its comprehensive approach, which includes the use of primary human cell lines and mouse models, and focuses on RNA-protein interactions and their effects on chromatin and gene expression.

• The authors present evidence that suggests that truncated versions of the CDKN1A interacting zinc finger protein 1, a protein coding gene, has major effects in the formation and stability of  chromatin. The presence of these truncated CIZ1 versions in cancer samples and the inability of large-scale transcriptome studies to detect this behavior, is striking, and adds value to the overall experimental rationale presented by this study. 

• The graphical abstract is a useful addition for comprehending the overarching interpretation of the data by the authors.

Major aspects to be addressed:

• In the first result section, the authors conclude that RD and AD must be in different polypeptides. We consider that this assumption is overreaching. Some considerations are the following:

  • It would be helpful to explain the purpose and expected result and/or interpretation of the high salt extraction in more detail since this experiment seems key to the conclusion they are trying to draw regarding the uncoupling of RD and AD staining

  • Figure 1F is not strong enough to convince that the RD and AD are differentially susceptible to extraction.We observe that in all cell lines the signal for both is decreased. We suggest quantifying the intensity of staining of both RD and AD in the nucleus with and without extraction would provide more convincing support to draw this conclusion.

  • We believe that a differential staining by the two antibodies in a single sample is, on its own, not sufficient to conclude that they are recognizing different polypeptides. The differential staining could also be explained by differential epitope accessibility.

• We find that the text is in many occasions ambiguous describing if an experiment was carried out on human or mouse samples. The text should clearly distinguish between experiments done in the human versus the mouse models. For example, we find not that it is not clear enough what is the overlap between the up-regulated genes observed in the human versus the mouse models.

• The overall conclusions reached by the authors heavily depend on the mouse model faithfully replicating the human model. The authors should discuss more in depth to what extent mouse large non-coding RNAs are homologous to human large non-coding RNAs (e.g., Xist lncRNA). Similarly, has the physical interaction between the human and mouse CIZ1 proteins has ever been demonstrated outside the murine model? 

Minor comments:

Regarding the introduction:

• To avoid confusing readers, the authors should present both terms Xi and Barr body together  and explain if they are interchangeable. 

• When introducing the general knowledge on CZ1 molecular biology the authors should include more citations than its own group. We would recommend finding additional sources.  

• We would like to encourage the authors to broaden the scope to incorporate other well-established epigenetic changes in breast cancer. Comparative studies or references to these processes will help contextualize the findings and strengthen the claim that CIZ1 is a significant factor in breast cancer epigenetics.

Regarding results and figures:

• In general we find that some figures, especially those showing disrupted CIZ1 assemblies, are difficult to interpret without clearer labeling or more descriptive legends. Additionally, some figures lack statistical annotations (e.g., p-values) to indicate the significance of the observed differences.

• In figure 1: 

  • it would be helpful to label the protein domains and/or motifs directly in this figure, rather than having to refer to the supplemental figure.

  • Statistics presented for the comparisons here would strengthen the authors' conclusions.

  • To strengthen the conclusions of the first result section, we would recommend including western blots.

• In figure 2: 

  • We would appreciate a more explicit explanation of how relative quantification was calculated here would be helpful.

  • The labeling of the genes is not uniform. The gene Estrogen Receptor 1, should be abbreviated just like the others are. In addition, if the text: "Normalized exon coverage" is supposed to apply to the three genes panel, then it must be vertically centered. As it is now, it only seems to apply to the lower panels.

  • The conclusions here would be strengthened by a similar analysis of CIZ1 exon expression in non-cancer cells or tissues.

  • We would suggest revising the number of transcriptomes submitted (10. The number of breast cancer samples presented in SFig.2C is 1085, not 1095. The authors need to either correct or explain the difference. 

  • Can the authors address why the coverage of the 5' exons (in particular exon 2) is so much lower than exon 7? The difference between exon 2 and exon 7 (~5-fold) is even greater than the increase of the 3' exons (~1.5-2 fold) that they are focused on.

  • We find the sentence: "While transcripts that map to the whole CIZ1 gene revealed no overall difference in expression between tumors and normal tissue (SFig.2C)," misleading. The normalized Transcripts Per Million (TPM) values are generated by mapping to the genome Next Generation Sequencing Read (NGS), and then normalizing these values. These numbers are not generated by mapping transcripts to the genome.

  • A more detailed explanation of how AD RNA levels of just one specific domain can be monitored through the transcriptomic studies, would be appreciated

• In the first paragraph of the third results subsection, it would be helpful for the reader to show or explain more explicitly the location of the RNA interaction surfaces relative to the other parts of the protein being referred to, i.e. AD vs RD, and the overrepresented AD described in cancers in previous section

• In the last paragraph of the third results subsection, it would be more clear if the types of assemblies previously defined were explicitly referred to in the text. The way I understand, you have defined CIZ1-Xi assemblies in D3T3 cells as:

Large discrete CIZ1 SMACs assemblies (Type 1),

Intermediate CIZ1 SMACs assemblies (Type 2)

No detectable CIZ1 SMACs assemblies (Type 3)

We would also like to suggest to the authors incorporate these definitions throughout the rest of the text, as and when needed.

• In the first paragraph of the forth results subsection, the abbreviation PRC should be properly defined, before being introduced into the text. 

• In figue 6, the text of this panel is not being displayed correctly. It seems to be covered by Panel A.

• Regarding Methods:

  • The manuscript lacks some comprehensive details, such as how tissue samples were selected and processed. While there is mention of the use of primary human mammary epithelial cells (HMECs), the selection criteria for the specific breast cancer cell lines and their detailed characteristics are not fully explained. Thus, authors should provide more details regarding the criteria for selecting cell lines, tissue samples, and the extent to which these models represent early-stage breast cancer. This can be supplemented by including a flowchart or table summarizing the experimental design.

  • Mouse transcriptomics, as this section is presented, it would be hard to replicate, without more computational details. We suggest to:

    • Provide the NCBI-SRA accession numbers for these datasets.

    • Describe the experimental design of the experiment and the number of experimental replicates.

    • Provide example commands for the different steps of the analysis.

    • In a table, please indicate the total number of reads obtained per sequencing library and the % of reads that were lost after Quality Control (QC). Also elaborate on how the low quality reads were removed.

  • Patient and Cell Line Bioinformatics: 

    • The exact command (and flags) used in the exclusion of secondary and supplementary alignments must be provided.

  • Analysis of domain imbalance

    • We would suggest the use of "non-coding exons" and "coding exons" as opposed to "translated exons" and "untranslated exons", as this is the most commonly used in computational genomics.

    • Please indicate how the alignments were generated, and indicate what outputs were normalized - those of the alignment or those of IGV. As written this paragraph is unclear.

  • ATACseq analysis

    • Please indicate the exact command (and flags) used in the exclusion of secondary and supplementary alignments.

    • Also the DESeq2 R code used should be provided.

• Supplemental material: We were unable to access the supplemental data sets Excel files. This should be corrected.

Regarding the discussion: 

• The manuscript mainly focuses on breast cancer but does not address whether the findings related to CIZ1 could be applicable to other types of cancer, especially those where chromatin instability plays a key role. We would encourage the authors to add a brief section or discussion on the potential implications of these findings for other cancers, drawing on existing literature about CIZ1's role in chromatin stability across different types of malignancies.

• The discussion does not sufficiently link the findings to clinical implications, particularly in terms of how this epigenetic instability driven by CIZ1 could be used as a biomarker or therapeutic target. We would suggest including a more in-depth discussion about the potential clinical applications of these findings, such as CIZ1 as a diagnostic marker or therapeutic target for early detection and personalized treatment. Consider addressing the challenges in translating this research into clinical practice.

• Author should also address the possible influence of confounding factors and how they were controlled or accounted for in the study. A discussion on limitations and future directions to explore these variables would also strengthen the manuscript.

Comments on reporting:

• The manuscript does not provide enough detail on how sample sizes were determined for the experiments. This lack of justification may raise concerns about the robustness of the conclusions drawn from the data. Including a section justifying the sample sizes and power analyses used in the study will strengthen the credibility of the results and address any concerns about data reliability.

• The authors need to provide evidence that the NGS data associated with this study has been deposited into the NCBI - SRA and that they have associated Bioproject and Biosamples identifiers.

• Similarly, the authors need to provide examples of the code and/or scripts used during the study. Generation of a public GitHub or Zenodo repository is highly recommended.

Suggestions for future studies:

• Investigate whether altering the glycosylation pattern at Asn180 could enhance antibody binding to the epitope and improve vaccine efficacy.

• Design vaccine antigens that focus on the identified cross-protomer epitope to potentially elicit a stronger immune response.

• Obtain high-resolution structures of the equine H3N8 HA protein in complex with neutralizing antibodies to confirm the binding sites and interactions.

• Compare the sequence and structure of the equine H3N8 HA with other influenza strains to identify additional conserved epitopes.

Conflicts of interest of reviewers:

Any reviewer declared conflicts of interest

Competing interests

The authors declare that they have no competing interests.