Menu

Close

AGO1x prevents dsRNA-induced interferon signaling to promote breast cancer cell proliferation

Souvik Ghosh, Joao C Guimaraes, Manuela Lanzafame, Alexander Schmidt, Afzal Pasha Syed, Beatrice Dimitriades, Anastasiya Börsch, Shreemoyee Ghosh, Ana Luisa Correia, Johannes Danner, Gunter Meister, Luigi M. Terracciano, Salvatore Piscuoglio, Mihaela Zavolan

Preprint posted on April 10, 2019 https://www.biorxiv.org/content/10.1101/603506v1

When do stop signs become optional? Ghosh et al. report an important role for an AGO1 isoform generated by translational readthrough.

Selected by Lorenzo Lafranchi

 

Background

Programmed and efficient stop codon readthrough has been initially observed in viruses, but its occurrence has recently been confirmed in other organisms, including humans. Translational readthrough diversifies the proteome by producing low abundant protein isoforms carrying C-terminal extensions. Often, the function of these isoforms differs from that of the canonical protein. The efficiency of translational readthrough depends on a variety of factors. For example, cis-acting elements located in 3’ untranslated region (3’ UTR) of mRNAs can facilitate stop codon readthrough. In the case of the vascular endothelial growth factor A (VEGF-A), binding of the heterogeneous nuclear ribonucleoprotein (hnRNP) A2/B1 protein to the 3’ UTR of the mRNA promotes synthesis of an extended isoform dubbed VEGF-Ax (Eswarappa et al, 2014). The same study identified the mRNA of Argonaute 1 (AGO1), a protein well known for its role in RNA-mediated post-transcriptional gene silencing, as a putative target of translational readthrough. However, a thorough characterization of the arising AGO1 isoform was missing so far.

 

Key findings

Corroborating the hypothesis that the Argonaute 1 gene is prone to translational readthrough, the authors observed that the region downstream of the AGO1 stop codon is highly conserved across vertebrates. Western blot analysis of AGO1 expression levels in different cell lines revealed a second band of higher molecular weight in addition to the main, expected AGO1 signal. Interestingly, the upper band is more prominent in a breast cancer cell line, compared to the two other cancer cell lines used. Based on these data, the authors speculated that an amplification of the AGO1 locus in breast and ovarian cancer xenografts results in higher translational readthrough of the AGO1 gene. To confirm that the slower-migrating band belongs to a readthrough-dependent AGO1 isoform, the authors raised an antibody directed to the readthrough region of the peptide. With this tool in hand, and supporting their finding with mass spectrometry, the authors convincingly show that translational readthrough takes place on the AGO1 transcript, resulting in the AGO1x isoform.

 

Unlike canonical AGO1, which resides in the cytoplasm, AGO1x is enriched around the nucleoli, suggesting a different functional role for this isoform. To better understand AGO1x function, two independent mutant cell lines lacking the AGO1x isoform were generated directing the CRISPR/Cas9 system at the predicted readthrough region. Characterizing the mutants, the authors noticed that in absence of AGO1x cell growth and mobility are reduced. Consistently, immunohistochemistry of breast cancer tissues revealed that AGO1x is principally expressed in highly proliferative cells. To define how AGO1x is supporting cell proliferation, the authors analyzed the transcriptome of control and mutant cell lines. RNA sequencing data showed that hundreds of genes are differentially-expressed in AGO1x-deficient cells and these changes are consistent for the two mutant lines. A detailed analysis revealed that expression of genes related to the interferon alpha response and to the apoptosis pathways are increased in the mutant cells relative to control. In line with this observation, treatment with ruxolitinib, an inhibitor of the interferon response fully rescued the growth defect of the mutant cell lines.

 

By testing their hypothesis that activation of the interferon response in the AGO1x-deficient cells could be due to the accumulation of cytoplasmic double-stranded RNA (dsRNA), the authors demonstrate that the levels of several intracellular dsRNA sensors are increased in the mutant cells compared to the parental cell line. As suggested by the increased amount of dsRNA sensors, higher levels of dsRNA, in particular belonging to rRNAs and GC-rich mRNAs, were detected in the absence of AGO1x. Finally, the Polyribonucleotide nucleotidyltransferase 1 (PNPT1) and the ATP-dependent RNA helicase A (DHX9) were identified as interaction partners of AGO1x. Altogether, these data suggest a role for this newly-discovered complex in sequestering and, possibly, degrading dsRNA. This protective mechanism seems to be particularly important for highly proliferative cells, relying on an increased transcription of ribosomal DNA.

 

What I like about this work and future direction

It has recently been proposed that cancer cells rely on an alternative translational program, skewed towards oncogenic mRNA translation and sustained translation of upstream open reading frames (uORF). Similarly, the data presented in this paper suggest that translational readthrough could be an additional feature of translational reprogramming of cancer cells. Nevertheless, more data are required to understand if translational readthrough in cancer cells is specifically directed to a subset of genes, as in the case of Argonaute 1, or is a widespread phenomenon.

 

Questions

The authors speculate that amplification of the AGO1 locus in breast and ovarian cancers results in higher expression of AGO1x. Why would this amplification not result in a concomitant increase of the canonical gene product?

Is it possible that increased levels of the heterogeneousribonucleoprotein A2/B1 in cancer cells (or generally in proliferative cells) are responsible for the increment in translational readthrough? Did you compare hnRNP A2/B1 levels between the cell lines used in this study?

Does depletion of hnRNP A2/B1 recapitulate the phenotypes observed in the AGO1x-deficient cell lines?

Is it possible to use the MS data presented in fig. 1e to define which amino acid is used to suppress the stop codon of the AGO1 gene?

 

References

Eswarappa SM, Potdar AA, Koch WJ, Fan Y, Vasu K, Lindner D, Willard B, Graham LM, DiCorleto PE, Fox PL. Programmed translational readthrough generates antiangiogenic VEGF-Ax. Cell. 2014 Jun 19;157(7):1605-18.

 

Posted on: 5th May 2019 , updated on: 7th May 2019

Read preprint (No Ratings Yet)




  • Author's response

    Joao Guimaraes and Mihaela Zavolan shared

    • The authors speculate that amplification of the AGO1 locus in breast and ovarian cancers results in higher expression of AGO1x. Why would this amplification not result in a concomitant increase of the canonical gene product?

    This is an interesting and likely complex question that we have not yet investigated. Indeed, if the readthrough is not specifically facilitated in the cancer cells at the expense of the canonical product, the canonical product should increase as well. Furthermore, there are reports that Ago proteins influence each other’s expression, and that the localization of Ago proteins varies between cell types and conditions. Overall, it seems that the functions of Ago family proteins have still not been fully charted.

    • Is it possible that increased levels of the heterogeneousribonucleoprotein A2/B1 in cancer cells (or generally in proliferative cells) are responsible for the increment in translational readthrough? Did you compare hnRNP A2/B1 levels between the cell lines used in this study?

    The AGO1 3’UTR is predicted to have a binding site for hnRNP A2/B1 and therefore this protein could regulate AGO1 stop codon readthrough, as it has been previously described for VEGF-Ax. In our study, we focused on the Ago1x function, about which nothing was known, rather than on the regulation of Ago1x expression. Clearly there is much more work to be done in this area, especially given the increased expression of Ago1x in the context of cancer cells. The mechanisms behind this expression pattern are an indeed what we would like to learn about in the immediate future.

    • Does depletion of hnRNP A2/B1 recapitulate the phenotypes observed in the AGO1x-deficient cell lines?

    As said, we have not tested whether hnRNP A2/B1 is the factor behind Ago1x expression in cancer cells. A few recent studies showed that inhibition of HNRNPA2B1 has potential antitumor effects in a pancreatic cancer cell model (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5223355/) and in glioblastoma (http://cancerres.aacrjournals.org/content/71/13/4464). Although these findings are consistent with the notion that HNRNPA2B1 regulates the translation readthrough of AGO1, this hypothesis needs to be directly tested.

    • Is it possible to use the MS data presented in fig. 1e to define which amino acid is used to suppress the stop codon of the AGO1 gene?

    We did wonder about this, but unfortunately, the peptide that covers the stop codon region is too short to be resolved by mass spectrometry. One would have to experiment with different digestion methods during sample preparation so that a peptide of appropriate size is generated. Clearly, there is much more to be learned about this peculiar and highly conserved isoform, and we hope that our findings will prompt further studies.

    Have your say

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.

    Sign up to customise the site to your preferences and to receive alerts

    Register here
    Close