To make a protein, the DNA sequence that encodes it is first copied to make a molecule of messenger RNA (or mRNA for short). The mRNA is then used as a set of instructions to assemble a protein in a process called translation. Both DNA and RNA molecules can fold into particular shapes. One such structure is known as a G-quartet and involves the DNA or RNA folding back on itself to form a highly stable planar structure. Stacks of G-quartets can form structures known as G-quadruplexes, but little is known about the G-quadruplexes that form in mRNA molecules.

Leukemia affects cells in the bone marrow and causes blood cells to develop abnormally. A protein called Aven is often found in increased amounts in certain types of leukemic cells, but it was not clear how Aven affects how leukemia develops. Thandapani, Song et al. have now found that in leukemic cells, Aven binds to G-quadruplexes found in two mRNA molecules that encode proteins that are linked to leukemia. This binding increases the translation of these mRNAs, with translation occurring most efficiently when a particular enzyme called a helicase—which remodels RNA—also bound to Aven. Reducing the amount of Aven in cells caused fewer of the leukemic proteins to be produced, which also reduced the growth and multiplcation of leukemic cells.

These findings raise the possibility that drugs that disrupt how Aven works could form part of treatments for leukemia. The next challenge will be to identify the signaling pathways that communicate with Aven and to define all the G-quadruplex mRNAs that are regulated by Aven.

RNA-binding proteins (RBPs) coordinate many steps of RNA metabolism ranging from splicing, RNA processing, RNA transport, mRNA translation, and RNA degradation (Glisovic et al., 2008). RBPs associate with specific RNA motifs and/or secondary structures within coding, untranslated regions, and non-coding RNAs in functional units called ribonucleoprotein (RNP) complexes (Mitchell and Parker, 2014). Defects in RBPs have been associated with many complex diseases ranging from neurological disorders to cancer (Lukong et al., 2008; Cooper et al., 2009; Castello et al., 2013; Ramaswami et al., 2013).

RBPs are predominantly defined by the presence of RNA-binding domains within their sequences (Chen and Varani, 2013). Recently, several ‘interactome capture’ strategies have been performed to identify RBPs genome-wide. In addition to identifying the known RBPs, these approaches have identified numerous mammalian proteins that do not possess a canonical RNA-binding domain (Baltz et al., 2012; Castello et al., 2012; Kwon et al., 2013). Interestingly, RBPs that harbor repeated sequences including YGG and RGG motifs were identified (Castello et al., 2012). The RGG/RG motif is enriched in proteins associated with RNA and is a known RNA-binding interface (reviewed in Thandapani et al., 2013). The RGG/RG motif, also called RGG box, was shown to bind RNA (Kiledjian and Dreyfuss, 1992). Subsequently, the RGG/RG motifs of Nucleolin, FMRP, FUS, and EWS were also shown to bind guanine-rich sequences that are potential G-quadruplexes (Darnell et al., 2001; Takahama et al., 2011, 2013; Haeusler et al., 2014). RGG/RG motifs also mediate protein–protein interactions. Notably, the RGG/RG motif of yeast Scd6 mediates interactions with eIF-4G, which leads to stress granule formation and inhibition of cap-dependent translation (Rajyaguru et al., 2012). Despite these recent advances, little is known about the role of RGG/RG motifs that bind both RNA and proteins.

G-quadruplexes (G4) are planar structures of stacks of guanine tetrads stabilized by monovalent potassium or sodium ions. G-quadruplexes have been shown to regulate DNA replication, DNA repair, gene expression, and telomeres (Bates et al., 2007). Less is known about G4 structures found in RNA. There are >1500 potential G4s (PG4s) in 5′-UTR of mRNAs alone (Beaudoin and Perreault, 2010), but not all PG4s form stable G-quadruplexes, which are influenced by the numbers of G-quartets, the possibility of bulge formation, the length of the loops, and the presence of alternative Watson–Crick base pair-based stable structure (Burge et al., 2006; Mukundan and Phan, 2013; Jodoin et al., 2014). Some PG4s in the 5′-UTR of mRNAs form bona fide G-quadruplexes and inhibit cap-dependent translation (Kumari et al., 2007; Beaudoin and Perreault, 2010; Bugaut and Balasubramanian, 2012). Recently, inhibitors of DEAD box RNA helicase eIF4A or eIF4A1 depletion have been shown to selectively inhibit translation of mRNAs with G-quadruplexes in their 5′ UTR (Wolfe et al., 2014; Modelska et al., 2015). However, presence of G-quadruplexes in 5′ UTRs does not appear to be sufficient to render translation of mRNAs sensitive to changes in eIF4A activity (Rubio et al., 2014). In addition to the incomplete understanding of the role of 5′ UTR G-quadruplexes in translation control, little is known about how G4 structures in open-reading frames (ORFs) affect translation.

Arginine residues within RGG/RG motifs are preferred substrates for methylation by protein arginine methyltransferases (PRMTs) (Thandapani et al., 2013). Arginine methylation is known to regulate many cellular processes including signal transduction, transcription, pre-mRNA splicing, and DNA repair (Bedford and Richard, 2005; Bedford and Clarke, 2009; Xu et al., 2013). PRMT1 generates >85% of asymmetric dimethylarginines found in cells with preference for RGG/RG motif containing proteins (Bedford and Clarke, 2009). PRMT1 is known for its nuclear roles in regulating gene expression and DNA damage (Strahl et al., 2001; Wang et al., 2001; An et al., 2004; Boisvert et al., 2005), however, less is known about its cytoplasmic roles. PRMT1-deficient mice die at E6.5 and the absolute removal of PRMT1 in mouse embryo fibroblasts (MEFs) leads to cell death (Pawlak et al., 2000; Yu et al., 2009). To identify other biological processes regulated by arginine methylation, we performed a bioinformatics approach to identify proteins harboring RGG/RG motifs and one such protein we identified was Aven (Thandapani et al., 2013).

Aven is a predominantly cytoplasmic protein required for cell survival and it has been shown to function as an apoptotic inhibitor by interaction with and stabilizing the pro-survival protein Bcl-x_L, as well as inhibiting the function of Apaf-1 (Chau et al., 2000). It was proposed that the proteolytic cleavage of Aven by Cathepsin D is required for its anti-apoptotic activity (Melzer et al., 2012). Furthermore, Aven is required for ataxia telangiectasia-mutated (ATM) activation in Xenopus oocytes and HeLa cells (Guo et al., 2008) and ataxia telangiectasia-related activation following DNA damage in osteosarcoma cells (Baranski et al., 2015). High Aven expression correlates with poor survival in metastatic patients with osteosarcomas (Baranski et al., 2015). The elevated Aven expression is also frequently observed in acute myeloid leukemia and acute lymphoblastic leukemia (T-ALL) and is associated with poor prognosis (Paydas et al., 2003; Choi et al., 2006). A transgenic mouse model with T cell-specific overexpression of Aven showed that its expression enhanced T-cell lymphomagenesis in the absence of p53 (Eismann et al., 2013). The mechanism by which Aven promotes hematological malignancies is yet to be understood.

Herein, we report that the methylation of the RGG/RG motif of Aven functions in the translational control of mRNAs harboring G4 structures in their ORFs. The association of Aven with polysomes was dependent on the arginine methylation of its RGG/RG motif and on the methyl-dependent interactions with the Tudor domains of SMN and TDRD3, previously shown to be associated with polysomes (Goulet et al., 2008; Sanchez et al., 2013). We identify Aven to be an RBP, as its RGG/RG motif bound G4 motifs in the ORFs of mRNAs encoding the mixed lineage leukemia (MLL) family proteins MLL1 and MLL4. The RGG/RG motif of Aven also associated with the G4 RNA helicase, DHX36, and this helicase was required for optimal translation of Aven-regulated mRNAs. Furthermore, Aven-deficient T-ALL cell lines had reduced MLL1 and MLL4 protein levels, but not mRNA levels, which were paralleled by proliferation defects. These findings define a hitherto unknown mechanism of action for arginine methylation in regulating translation of a subset of mRNAs including those encoding pivotal leukemogenic transcriptional regulators MLL1 and MLL4.

The MLL proto-oncogene encodes a histone methyltransferase implicated in epigenetic modifications, regulating gene expression for embryonic development and hematopoiesis (Liedtke and Cleary, 2009; Smith et al., 2011). MLL is a recurrent site of DNA translocations resulting in an MLL fusion protein where the N-terminus of the MLL is fused to a variety of proteins (Liedtke and Cleary, 2009; Smith et al., 2011). In the present manuscript, we identify, within the mRNA-coding regions of MLL1 and MLL4, RNA elements that regulate its polysomal association and protein synthesis. These RNA elements are located between 200 and 300 nucleotides downstream of the initiator methionine ATG and encode protein sequences rich in glycines and arginine–glycine repeats in MLL1 and MLL4. The function of these N-terminal repeats is unknown. We show that Aven binds the MLL1 and MLL4 G4 RNA structures in vitro and in vivo with its RGG/RG motif. Aven was required for the translational regulation of MLL1 and MLL4, as Aven-deficient T-ALL cells exhibited decreased MLL1 and MLL4 protein expression and consequently decreased the expression of their downstream targets including, the HOX genes. The association of Aven with polysomes required the methylation of its RGG/RG motif by PRMT1 and interaction with methyl-binding proteins, TDRD3 and SMN. The Aven interaction with TDRD3 and SMN may require other protein or RNA components in the complex for enhanced association. Deficiency of Aven or PRMT1 in acute leukemic cell lines led to decreased cell proliferation. Taken together, our studies suggest that Aven regulates the translation of MLL1 and MLL4 required for cancer survival and that targeting this pathway may have therapeutic potential.

RGG/RG motifs have the biochemical properties to bind both RNA and proteins to fulfill their emerging roles in assembly of RNP complexes and translational control (Rajyaguru and Parker, 2012; Thandapani et al., 2013). The RGG/RG motif of yeast proteins Scd6, Npl3, and Sbp1 was shown to interact with the translational initiation factor eIF4G and repress translation by preventing the formation of pre-initiation complex (Rajyaguru et al., 2012). In trypanosomes, the RGG/RG motif of SCD6 is involved in regulating the type and number of RNP granules (Krüger et al., 2013). Amyloid-like fibers were formed when the RGG/RG motif of FUS was incubated with RNA (Schwartz et al., 2013). These fibers are characterized by the reversible transformation from soluble to polymeric state (Han et al., 2012; Kato et al., 2012). Although many proteins have an RGG/RG motif, the Aven RGG/RG motif may be more accessible, as it is located at the N-terminus and may protrude outwards. Since the Aven RGG/RG motif is not required for self-association (Figure 2—figure supplement 1), this suggests that an Aven dimer has 2 protruding RGG/RG motifs that can each mediate their own interactions. Therefore, we speculate that Aven functions as a scaffolding protein to assemble translationally competent RNPs for certain mRNAs containing G4 motifs (Figure 11).

Figure 11

Download asset Open asset

DHX36 is a DEAH (aspartic acid, glutamic acid, alanine, histidine)-box helicase and it is the only G4 RNA resolvase known and is a major DNA G4 resolvase (Creacy et al., 2008; Lattmann et al., 2010). Aven associated with DHX36 to regulate translation of mRNAs with G4 structures. DHX36 knockdown increased the expression of PITX1 protein without changes in mRNA, suggesting that it functions in translational control (Booy et al., 2014). Ribosomal footprinting studies have led to the proposal that elongating ribosomes likely use accessory RNA helicases (Rouskin et al., 2014), and our data suggest that DHX36 may be such an accessory helicase. DHX36 null mice are embryonic lethal and deletion in the hematopoietic system using Vav1-Cre causes hemolytic anemia and defects at the proerythroblast stage with deregulation of genes with G4 motifs in their promoters (Lai et al., 2012), however, a role DHX36 in translational control was not examined.

Many PRMT1 substrates are RBPs with RGG/RG motifs (Bedford and Clarke, 2009) and some have been shown to associate with RNAs with G4 motifs such as Nucleolin, FUS, EWS, and FMRP (Thandapani et al., 2013). This suggests that several RBPs likely function in a similar manner to Aven in regulating accessibility of mRNPs with polysomes. It has been shown that the RGG/RG motif of FMRP is required for its polysomal association (Blackwell et al., 2010), however, whether arginine methylation by PRMT1 regulates association is unknown. Our findings show for the first time that arginine methylation by PRMT1 regulates translational control. It is known, however, that the yeast homolog of PRMT3 (RMT3) methylates rpS2, regulating the balance between the small and large ribosomal subunits (Bachand and Silver, 2004). However, mammalian PRMT3 did not influence ribosomal assembly or polysomal formation (Swiercz et al., 2004).

It is known that secondary RNA structure including G-quadruplex structures within mRNAs hinder mRNA translation (Koromilas et al., 1992; Sonenberg and Hinnebusch, 2009). Stable RNA secondary structures within the 5′-UTRs of mRNAs reduce cap-dependent translation by preventing assembly of the translational initiation machinery at the 5′-cap and also impair the scanning of the start site AUG by the initiation complex (Beaudoin and Perreault, 2010; Bugaut and Balasubramanian, 2012). Secondary structure in the 5′ UTRs including G4 motifs has been shown to require eIF4A for optimal translation output (Wolfe et al., 2014). The 5′-UTR of NRAS and Zic-1, which harbor G4 structures, reduce translation of a reporter luciferase (Kumari et al., 2007; Arora et al., 2008). G-quadruplex structures within ORFs of the virally encoded EBNA1 transcript were shown to hinder translational elongation by either ribosomal pausing or ribosomal dissociation (Murat et al., 2014). We now extend these observations and identify a mechanism regulated by arginine methylation that leads to the positive regulation of mRNAs with G4 structures within their coding region.

Aven is overexpressed in acute leukemia and was proposed to be a prognostic factor in acute childhood lymphoblastic leukemia for poor outcome (Choi et al., 2006). Aven is a well-established cell survival protein or inhibitor of apoptosis that prevents apoptosis by stabilizing pro-survival protein Bcl-x_L and inhibiting the function of pro-apoptotic protein Apaf-1 (Chau et al., 2000; Kutuk et al., 2010). It was reported that an N-terminal deleted fragment of Aven cleaved by cathepsin D harbors its anti-apoptotic function (Melzer et al., 2012), however, such a ∼30 kDa Aven species was not visible in MOLT4 and CCRF-CEM cells (Figure 6A,B) and was faintly observed in HEK293T (Figure 7A) as previously described (Melzer et al., 2012). Thus, cathepsin D mediated cleavage of Aven is unlikely involved in the regulation of translational control described herein. In addition to its pro-survival functions, Aven was identified to be essential for progression of acute leukemia in mice (Eismann et al., 2013). The regions required for association with Bcl-x_L and Apaf-1 reside C-terminal of the RGG/RG motif. Taken together with our findings, this suggests that Aven uses several mechanisms to increase cell survival, (1) preventing apoptosis via Bcl-x_L and Apaf-1, and (2) favoring the translation of mRNAs, including those encoding MLL1 and MLL4 required for cell survival.

PRMT1 was shown to be essential for MLL by the MLL-EEN gene fusion protein (Cheung et al., 2007). The EEN fusion partner leads to the recruitment of PRMT1 to methylate histones and lead to gene activation (Cheung et al., 2007). Our findings identify a new role for PRMT1 in the cytoplasm that is required for cancer cell survival. This pathway is amenable to therapeutic intervention with future PRMT1 inhibitors and specific RNA G-quadruplex ligands.

1. 1. An W
   2. Kim J
   3. Roeder RG

   (2004) Ordered cooperative functions of PRMT1, p300, and CARM1 in transcriptional activation by p53

   Cell 117:735–748.

   https://doi.org/10.1016/j.cell.2004.05.009

   • Google Scholar
2. 1. Arora A
   2. Dutkiewicz M
   3. Scaria V
   4. Hariharan M
   5. Maiti S
   6. Kurreck J

   (2008) Inhibition of translation in living eukaryotic cells by an RNA G-quadruplex motif

   RNA 14:1290–1296.

   https://doi.org/10.1261/rna.1001708

   • Google Scholar
3. 1. Bachand F
   2. Silver PA

   (2004) PRMT3 is a ribosomal protein methyltransferase that affects the cellular levels of ribosomal subunits

   The EMBO Journal 23:2641–2650.

   https://doi.org/10.1038/sj.emboj.7600265

   • Google Scholar
4. 1. Baltz AG
   2. Munschauer M
   3. Schwanhäusser B
   4. Vasile A
   5. Murakawa Y
   6. Schueler M
   7. Youngs N
   8. Penfold-Brown D
   9. Drew K
   10. Milek M
   11. Wyler E
   12. Bonneau R
   13. Selbach M
   14. Dieterich C
   15. Landthaler M

   (2012) The mRNA-bound proteome and its global occupancy profile on protein-coding transcripts

   Molecular Cell 46:674–690.

   https://doi.org/10.1016/j.molcel.2012.05.021

   • Google Scholar
5. 1. Baranski Z
   2. Booij TH
   3. Cleton-Jansen AM
   4. Price LS
   5. van de Water B
   6. Bovée JV
   7. Hogendoorn PC
   8. Danen EH

   (2015) Aven-mediated checkpoint kinase control regulates proliferation and resistance to chemotherapy in conventional osteosarcoma

   Journal of Pathology.

   https://doi.org/10.1002/path.4528

   • Google Scholar
6. 1. Bates P
   2. Mergny JL
   3. Yang D

   (2007) Quartets in g-major. The first international meeting on quadruplex DNA

   EMBO Reports 8:1003–1010.

   https://doi.org/10.1038/sj.embor.7401073

   • Google Scholar
7. 1. Beaudoin JD
   2. Jodoin R
   3. Perreault JP

   (2013) In-line probing of RNA G-quadruplexes

   Methods 64:79–87.

   https://doi.org/10.1016/j.ymeth.2013.02.017

   • Google Scholar
8. 1. Beaudoin JD
   2. Jodoin R
   3. Perreault JP

   (2014) New scoring system to identify RNA G-quadruplex folding

   Nucleic Acids Research 42:1209–1223.

   https://doi.org/10.1093/nar/gkt904

   • Google Scholar
9. 1. Beaudoin JD
   2. Perreault JP

   (2010) 5′-UTR G-quadruplex structures acting as translational repressors

   Nucleic Acids Research 38:7022–7036.

   https://doi.org/10.1093/nar/gkq557

   • Google Scholar
10. 1. Beaudoin JD
    2. Perreault JP

    (2013) Exploring mRNA 3′-UTR G-quadruplexes: evidence of roles in both alternative polyadenylation and mRNA shortening

    Nucleic Acids Research 41:5898–5911.

    https://doi.org/10.1093/nar/gkt265

    • Google Scholar
11. 1. Bedford MT
    2. Clarke SG

    (2009) Protein arginine methylation in mammals: who, what, and why

    Molecular Cell 33:1–13.

    https://doi.org/10.1016/j.molcel.2008.12.013

    • Google Scholar
12. 1. Bedford MT
    2. Richard S

    (2005) Arginine methylation an emerging regulator of protein function

    Molecular Cell 18:263–272.

    https://doi.org/10.1016/j.molcel.2005.04.003

    • Google Scholar
13. 1. Blackwell E
    2. Zhang X
    3. Ceman S

    (2010) Arginines of the RGG box regulate FMRP association with polyribosomes and mRNA

    Human Molecular Genetics 19:1314–1323.

    https://doi.org/10.1093/hmg/ddq007

    • Google Scholar
14. 1. Blagoev B
    2. Kratchmarova I
    3. Ong SE
    4. Nielsen M
    5. Foster LJ
    6. Mann M

    (2003) A proteomics strategy to elucidate functional protein-protein interactions applied to EGF signaling

    Nature Biotechnology 21:315–318.

    https://doi.org/10.1038/nbt790

    • Google Scholar
15. 1. Blobel G
    2. Sabatini D

    (1971) Dissociation of mammalian polyribosomes into subunits by puromycin

    Proceedings of the National Academy of Sciences of USA 68:390–394.

    https://doi.org/10.1073/pnas.68.2.390

    • Google Scholar
16. 1. Boisvert FM
    2. Déry U
    3. Masson JY
    4. Richard S

    (2005) Arginine methylation of MRE11 by PRMT1 is required for DNA damage checkpoint control

    Genes & Development 19:671–676.

    https://doi.org/10.1101/gad.1279805

    • Google Scholar
17. 1. Boisvert FM
    2. Ahmad Y
    3. Gierliński M
    4. Charrière F
    5. Lamont D
    6. Scott M
    7. Barton G
    8. Lamond AI

    (2012)

    A quantitative spatial proteomics analysis of proteome turnover in human cells

    Molecular & Cellular Proteomics, 22, M111.011429, 10.1074/mcp.M111.011429.

    • Google Scholar
18. 1. Booy EP
    2. Howard R
    3. Marushchak O
    4. Ariyo EO
    5. Meier M
    6. Novakowski SK
    7. Deo SR
    8. Dzananovic E
    9. Stetefeld J
    10. McKenna S

    (2014) The RNA helicase RHAU (DHX36) suppresses expression of the transcription factor PITX1

    Nucleic Acids Research 42:3346–3361.

    https://doi.org/10.1093/nar/gkt1340

    • Google Scholar
19. 1. Booy EP
    2. Meier M
    3. Okun N
    4. Novakowski SK
    5. Xiong S
    6. Stetefeld J
    7. McKenna SA

    (2012) The RNA helicase RHAU (DHX36) unwinds a G4-quadruplex in human telomerase RNA and promotes the formation of the P1 helix template boundary

    Nucleic Acids Research 40:4110–4124.

    https://doi.org/10.1093/nar/gkr1306

    • Google Scholar
20. 1. Bugaut A
    2. Balasubramanian S

    (2012) 5′-UTR RNA G-quadruplexes: translation regulation and targeting

    Nucleic Acids Research 40:4727–4741.

    https://doi.org/10.1093/nar/gks068

    • Google Scholar
21. 1. Burge S
    2. Parkinson GN
    3. Hazel P
    4. Todd AK
    5. Neidle S

    (2006) Quadruplex DNA: sequence, topology and structure

    Nucleic Acids Research 34:5402–5415.

    https://doi.org/10.1093/nar/gkl655

    • Google Scholar
22. 1. Cahill NM
    2. Friend K
    3. Speckmann W
    4. Li ZH
    5. Terns RM
    6. Terns MP
    7. Steitz JA

    (2002) Site-specific cross-linking analyses reveal an asymmetric protein distribution for a box C/D snoRNP

    The EMBO Journal 21:3816–3828.

    https://doi.org/10.1093/emboj/cdf376

    • Google Scholar
23. 1. Castello A
    2. Fischer B
    3. Eichelbaum K
    4. Horos R
    5. Beckmann BM
    6. Strein C
    7. Davey NE
    8. Humphreys DT
    9. Preiss T
    10. Steinmetz LM
    11. Krijgsveld J
    12. Hentze MW

    (2012) Insights into RNA biology from an atlas of mammalian mRNA-binding proteins

    Cell 149:1393–1406.

    https://doi.org/10.1016/j.cell.2012.04.031

    • Google Scholar
24. 1. Castello A
    2. Fischer B
    3. Hentze MW
    4. Preiss T

    (2013) RNA-binding proteins in Mendelian disease

    Trends in Genetics 29:318–327.

    https://doi.org/10.1016/j.tig.2013.01.004

    • Google Scholar
25. 1. Chau BN
    2. Cheng EH
    3. Kerr DA
    4. Hardwick JM

    (2000) Aven, a novel inhibitor of caspase activation, binds Bcl-xL and Apaf-1

    Molecular Cell 6:31–40.

    https://doi.org/10.1016/S1097-2765(05)00021-3

    • Google Scholar
26. 1. Chen C
    2. Nott TJ
    3. Jin J
    4. Pawson T

    (2011) Deciphering arginine methylation: tudor tells the tale

    Nature Reviews. Molecular Cell Biology 12:629–642.

    https://doi.org/10.1038/nrm3185

    • Google Scholar
27. 1. Chen Y
    2. Varani G

    (2013) Engineering RNA-binding proteins for biology

    The FEBS Journal 280:3734–3754.

    https://doi.org/10.1111/febs.12375

    • Google Scholar
28. 1. Cheung N
    2. Chan LC
    3. Thompson A
    4. Cleary ML
    5. So CW

    (2007) Protein arginine-methyltransferase-dependent oncogenesis

    Nature Cell Biology 9:1208–1215.

    https://doi.org/10.1038/ncb1642

    • Google Scholar
29. 1. Choi J
    2. Hwang YK
    3. Sung KW
    4. Kim DH
    5. Yoo KH
    6. Jung HL
    7. Koo HH

    (2006) Aven overexpression: association with poor prognosis in childhood acute lymphoblastic leukemia

    Leukemia Research 30:1019–1025.

    https://doi.org/10.1016/j.leukres.2005.11.001

    • Google Scholar
30. 1. Cooper TA
    2. Wan L
    3. Dreyfuss G

    (2009) RNA and disease

    Cell 136:777–793.

    https://doi.org/10.1016/j.cell.2009.02.011

    • Google Scholar
31. 1. Cote J
    2. Richard S

    (2005) Tudor domains bind symmetrical dimethylated arginines

    The Journal of Biological Chemistry 280:28476–28483.

    https://doi.org/10.1074/jbc.M414328200

    • Google Scholar
32. 1. Cox J
    2. Mann M

    (2008) MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification

    Nature Biotechnology 26:1367–1372.

    https://doi.org/10.1038/nbt.1511

    • Google Scholar
33. 1. Creacy SD
    2. Routh ED
    3. Iwamoto F
    4. Nagamine Y
    5. Akman SA
    6. Vaughn JP

    (2008) G4 resolvase 1 binds both DNA and RNA tetramolecular quadruplex with high affinity and is the major source of tetramolecular quadruplex G4-DNA and G4-RNA resolving activity in HeLa cell lysates

    The Journal of Biological Chemistry 283:34626–34634.

    https://doi.org/10.1074/jbc.M806277200

    • Google Scholar
34. 1. Darnell JC
    2. Jensen KB
    3. Jin P
    4. Brown V
    5. Warren ST
    6. Darnell RB

    (2001) Fragile X mental retardation protein targets G quartet mRNAs important for neuronal function

    Cell 107:489–499.

    https://doi.org/10.1016/S0092-8674(01)00566-9

    • Google Scholar
35. 1. Eismann M
    2. Melzer IM
    3. Fernandez SBM
    4. Michel G
    5. Hrabe de Angelis M
    6. Hoefler G
    7. Finkenwirth P
    8. Jauch A
    9. Schoell B
    10. Grez M
    11. Schmidt M
    12. Bartholomae CC
    13. Newrzela S
    14. Haetscher N
    15. Rieger MA
    16. Zachskorn C
    17. Mittelbronn M
    18. Zörnig M

    (2013)

    Overexpression of the anti-apoptotic protein AVEN contributes to increased malignancy in hematopoietic neoplasms

    Oncogene 32:2586–2591.

    • Google Scholar
36. 1. Esmaili AM
    2. Johnson EL
    3. Thaivalappil SS
    4. Kuhn HM
    5. Kornbluth S
    6. Irusta P

    (2010) Regulation of the ATM-activator protein Aven by CRM1-dependent nuclear export

    Cell Cycle 9:3913–3920.

    https://doi.org/10.4161/cc.9.19.13138

    • Google Scholar
37. 1. Gandin V
    2. Gutierrez GJ
    3. Brill L
    4. Varsano T
    5. Feng Y
    6. Aza-Blanc P
    7. Au Q
    8. McLaughlan S
    9. Ferreira TA
    10. Alain T
    11. Sonenberg N
    12. Topisirovic I
    13. Ronai ZA

    (2013) Degradation of newly synthesized polypeptides by ribosome-associated RACK1/c-Jun N-terminal kinase/eukaryotic elongation factor 1A2 complex

    Molecular and Cellular Biology 33:2510–2526.

    https://doi.org/10.1128/MCB.01362-12

    • Google Scholar
38. 1. Gandin V
    2. Sikström K
    3. Alain T
    4. Morita M
    5. McLaughlan S
    6. Larsson O
    7. Topisirovic I

    (2014) Polysome fractionation and analysis of mammalian translatomes on a genome-wide scale

    Journal of Visualized Experiments.

    https://doi.org/10.3791/51455

    • Google Scholar
39. 1. Glisovic T
    2. Bachorik JL
    3. Yong J
    4. Dreyfuss G

    (2008) RNA-binding proteins and post-transcriptional gene regulation

    FEBS Letters 582:1977–1986.

    https://doi.org/10.1016/j.febslet.2008.03.004

    • Google Scholar
40. 1. Goulet I
    2. Boisvenue S
    3. Mokas S
    4. Mazroui R
    5. Côté J

    (2008) TDRD3, a novel Tudor domain-containing protein, localizes to cytoplasmic stress granules

    Human Molecular Genetics 17:3055–3074.

    https://doi.org/10.1093/hmg/ddn203

    • Google Scholar
41. 1. Guo JY
    2. Yamada A
    3. Kajino T
    4. Wu JQ
    5. Tang W
    6. Freel CD
    7. Feng J
    8. Chau BN
    9. Wang MZ
    10. Margolis SS
    11. Yoo HY
    12. Wang XF
    13. Dunphy WG
    14. Irusta PM
    15. Hardwick JM
    16. Kornbluth S

    (2008) Aven-dependent activation of ATM following DNA damage

    Current Biology 18:933–942.

    https://doi.org/10.1016/j.cub.2008.05.045

    • Google Scholar
42. 1. Haeusler AR
    2. Donnelly CJ
    3. Periz G
    4. Simko EA
    5. Shaw PG
    6. Kim MS
    7. Maragakis NJ
    8. Troncoso JC
    9. Pandey A
    10. Sattler R
    11. Rothstein JD
    12. Wang J

    (2014) C9orf72 nucleotide repeat structures initiate molecular cascades of disease

    Nature 507:195–200.

    https://doi.org/10.1038/nature13124

    • Google Scholar
43. 1. Han TW
    2. Kato M
    3. Xie S
    4. Wu LC
    5. Mirzaei H
    6. Pei J
    7. Chen M
    8. Xie Y
    9. Allen J
    10. Xiao G
    11. McKnight SL

    (2012) Cell-free formation of RNA granules: bound RNAs identify features and components of cellular assemblies

    Cell 149:768–779.

    https://doi.org/10.1016/j.cell.2012.04.016

    • Google Scholar
44. 1. Hawley RG
    2. Chen Y
    3. Riz I
    4. Zeng C

    (2012) An integrated bioinformatics and computational biology approach identifies new BH3-only protein candidates

    The Open Biology Journal 5:6–16.

    https://doi.org/10.2174/1874196701205010006

    • Google Scholar
45. 1. Huppertz I
    2. Attig J
    3. D'Ambrogio A
    4. Easton LE
    5. Sibley CR
    6. Sugimoto Y
    7. Tajnik M
    8. König J
    9. Ule J

    (2014) iCLIP: protein-RNA interactions at nucleotide resolution

    Methods 65:274–287.

    https://doi.org/10.1016/j.ymeth.2013.10.011

    • Google Scholar
46. 1. Jodoin R
    2. Bauer L
    3. Garant JM
    4. Mahdi Laaref A
    5. Phaneuf F
    6. Perreault JP

    (2014) The folding of 5′-UTR human G-quadruplexes possessing a long central loop

    RNA 20:1129–1141.

    https://doi.org/10.1261/rna.044578.114

    • Google Scholar
47. 1. Kato M
    2. Han TW
    3. Xie S
    4. Shi K
    5. Du X
    6. Wu LC
    7. Mirzaei H
    8. Goldsmith EJ
    9. Longgood J
    10. Pei J
    11. Grishin NV
    12. Frantz DE
    13. Schneider JW
    14. Chen S
    15. Li L
    16. Sawaya MR
    17. Eisenberg D
    18. Tycko R
    19. McKnight SL

    (2012) Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels

    Cell 149:753–767.

    https://doi.org/10.1016/j.cell.2012.04.017

    • Google Scholar
48. 1. Kiledjian M
    2. Dreyfuss G

    (1992)

    Primary structure and binding activity of the hnRNP U protein: binding RNA through RGG box

    The EMBO Journal 11:2655–2664.

    • Google Scholar
49. 1. Koromilas AE
    2. Lazaris-Karatzas A
    3. Sonenberg N

    (1992)

    mRNAs containing extensive secondary structure in their 5′ non-coding region translate efficiently in cells overexpressing initiation factor eIF-4E

    The EMBO Journal 11:4153–4158.

    • Google Scholar
50. 1. Krivtsov AV
    2. Armstrong SA

    (2007) MLL translocations, histone modifications and leukaemia stem-cell development

    Nature Rreviews. Cancer 7:823–833.

    https://doi.org/10.1038/nrc2253

    • Google Scholar
51. 1. Krüger T
    2. Hofweber M
    3. Kramer S

    (2013)

    SCD6 induces ribonucleoprotein granule formation in trypanosomes in a translation-independent manner, regulated by its Lsm and RGG domains

    Molecular Biology of the Cell 24:2098–2111.

    • Google Scholar
52. 1. Kumari S
    2. Bugaut A
    3. Huppert JL
    4. Balasubramanian S

    (2007) An RNA G-quadruplex in the 5′ UTR of the NRAS proto-oncogene modulates translation

    Nature Chemical Biology 3:218–221.

    https://doi.org/10.1038/nchembio864

    • Google Scholar
53. 1. Kutuk O
    2. Temel SG
    3. Tolunay S
    4. Basaga H

    (2010) Aven blocks DNA damage-induced apoptosis by stabilising Bcl-xL

    European Journal of Cancer 46:2494–2505.

    https://doi.org/10.1016/j.ejca.2010.06.011

    • Google Scholar
54. 1. Kwon SC
    2. Yi H
    3. Eichelbaum K
    4. Föhr S
    5. Fischer B
    6. You KT
    7. Castello A
    8. Krijgsveld J
    9. Hentze MW
    10. Kim VN

    (2013) The RNA-binding protein repertoire of embryonic stem cells

    Nature Structural & Molecular Biology 20:1122–1130.

    https://doi.org/10.1038/nsmb.2638

    • Google Scholar
55. 1. Lai JC
    2. Ponti S
    3. Pan D
    4. Kohler H
    5. Skoda RC
    6. Matthias P
    7. Nagamine Y

    (2012) The DEAH-box helicase RHAU is an essential gene and critical for mouse hematopoiesis

    Blood 119:4291–4300.

    https://doi.org/10.1182/blood-2011-08-362954

    • Google Scholar
56. 1. Lattmann S
    2. Giri B
    3. Vaughn JP
    4. Akman SA
    5. Nagamine Y

    (2010) Role of the amino terminal RHAU-specific motif in the recognition and resolution of guanine quadruplex-RNA by the DEAH-box RNA helicase RHAU

    Nucleic Acids Research 38:6219–6233.

    https://doi.org/10.1093/nar/gkq372

    • Google Scholar
57. 1. Liedtke M
    2. Cleary ML

    (2009) Therapeutic targeting of MLL

    Blood 113:6061–6068.

    https://doi.org/10.1182/blood-2008-12-197061

    • Google Scholar
58. 1. Lorenz R
    2. Bernhart SH
    3. Qin J
    4. Höner zu Siederdissen C
    5. Tanzer A
    6. Amman F
    7. Hofacker IL
    8. Stadler PF

    (2013) 2D meets 4G: G-quadruplexes in RNA secondary structure prediction

    IEEE/ACM Transactions on Computational Biology and Bioinformatics 10:832–844.

    https://doi.org/10.1109/TCBB.2013.7

    • Google Scholar
59. 1. Lukong KE
    2. Chang KW
    3. Khandjian EW
    4. Richard S

    (2008) RNA-binding proteins in human genetic disease

    Trends in Genetics 24:416–425.

    https://doi.org/10.1016/j.tig.2008.05.004

    • Google Scholar
60. 1. Mali P
    2. Yang L
    3. Esvelt KM
    4. Aach J
    5. Guell M
    6. DiCarlo JE
    7. Norville JE
    8. Church GM

    (2013) RNA-guided human genome engineering via Cas9

    Science 339:823–826.

    https://doi.org/10.1126/science.1232033

    • Google Scholar
61. 1. Melzer IM
    2. Fernández SB
    3. Bösser S
    4. Lohrig K
    5. Lewandrowski U
    6. Wolters D
    7. Kehrloesser S
    8. Brezniceanu ML
    9. Theos AC
    10. Irusta PM
    11. Impens F
    12. Gevaert K
    13. Zörnig M

    (2012) The Apaf-1-binding protein Aven is cleaved by Cathepsin D to unleash its anti-apoptotic potential

    Cell Death and Differentiation 19:1435–1445.

    https://doi.org/10.1038/cdd.2012.17

    • Google Scholar
62. 1. Mitchell SF
    2. Parker R

    (2014) Principles and properties of eukaryotic mRNPs

    Molecular Cell 54:547–558.

    https://doi.org/10.1016/j.molcel.2014.04.033

    • Google Scholar
63. 1. Modelska A
    2. Turro E
    3. Russell R
    4. Beaton J
    5. Sbarrato T
    6. Spriggs K
    7. Miller J
    8. Gräf S
    9. Provenzano E
    10. Blows F
    11. Pharoah P
    12. Caldas C
    13. Le Quesne J

    (2015) The malignant phenotype in breast cancer is driven by eIF4A1-mediated changes in the translational landscape

    Cell Death & Disease 6:e1603.

    https://doi.org/10.1038/cddis.2014.542

    • Google Scholar
64. 1. Mukundan VT
    2. Phan AT

    (2013) Bulges in G-quadruplexes: broadening the definition of G-quadruplex-forming sequences

    Journal of the American Chemical Society 135:5017–5028.

    https://doi.org/10.1021/ja310251r

    • Google Scholar
65. 1. Murat P
    2. Zhong J
    3. Lekieffre L
    4. Cowieson NP
    5. Clancy JL
    6. Preiss T
    7. Balasubramanian S
    8. Khanna R
    9. Tellam J

    (2014) G-quadruplexes regulate Epstein-Barr virus-encoded nuclear antigen 1 mRNA translation

    Nature Chemical Biology 10:358–364.

    https://doi.org/10.1038/nchembio.1479

    • Google Scholar
66. 1. Pawlak MR
    2. Scherer CA
    3. Chen J
    4. Roshon MJ
    5. Ruley HE

    (2000) Arginine n-methyltransferase 1 is required for early postimplantation mouse development, but cells deficient in the enzyme are viable

    Molecular and Cellular Biology 20:4859–4869.

    https://doi.org/10.1128/MCB.20.13.4859-4869.2000

    • Google Scholar
67. 1. Paydas S
    2. Tanriverdi K
    3. Yavuz S
    4. Disel U
    5. Sahin B
    6. Burgut R

    (2003) Survivin and aven: two distinct antiapoptotic signals in acute leukemias

    Annals of Oncology 14:1045–1050.

    https://doi.org/10.1093/annonc/mdg277

    • Google Scholar
68. 1. Phan AT
    2. Kuryavyi V
    3. Darnell JC
    4. Serganov A
    5. Majumdar A
    6. Ilin S
    7. Raslin T
    8. Polonskaia A
    9. Chen C
    10. Clain D
    11. Darnell RB
    12. Patel DJ

    (2011) Structure-function studies of FMRP RGG peptide recognition of an RNA duplex-quadruplex junction

    Nature Structural & Molecular Biology 18:796–804.

    https://doi.org/10.1038/nsmb.2064

    • Google Scholar
69. 1. Piques M
    2. Schulze WX
    3. Höhne M
    4. Usadel B
    5. Gibon Y
    6. Rohwer J
    7. Stitt M

    (2009) Ribosome and transcript copy numbers, polysome occupancy and enzyme dynamics in Arabidopsis

    Molecular Systems Biology 5:314.

    https://doi.org/10.1038/msb.2009.68

    • Google Scholar
70. 1. Rajyaguru P
    2. Parker R

    (2012) RGG motif proteins: modulators of mRNA functional states

    Cell Cycle 11:2594–2599.

    https://doi.org/10.4161/cc.20716

    • Google Scholar
71. 1. Rajyaguru P
    2. She M
    3. Parker R

    (2012) Scd6 targets eIF4G to repress translation: RGG motif proteins as a class of eIF4G-binding proteins

    Molecular Cell 45:244–254.

    https://doi.org/10.1016/j.molcel.2011.11.026

    • Google Scholar
72. 1. Ramaswami M
    2. Taylor JP
    3. Parker R

    (2013) Altered ribostasis: RNA-protein granules in degenerative disorders

    Cell 154:727–736.

    https://doi.org/10.1016/j.cell.2013.07.038

    • Google Scholar
73. 1. Rouskin S
    2. Zubradt M
    3. Washietl S
    4. Kellis M
    5. Weissman JS

    (2014) Genome-wide probing of RNA structure reveals active unfolding of mRNA structures in vivo

    Nature 505:701–705.

    https://doi.org/10.1038/nature12894

    • Google Scholar
74. 1. Rubio CA
    2. Weisburd B
    3. Holderfield M
    4. Arias C
    5. Fang E
    6. DeRisi JL
    7. Fanidi A

    (2014) Transcriptome-wide characterization of the eIF4A signature highlights plasticity in translation regulation

    Genome Biology 15:476.

    https://doi.org/10.1186/s13059-014-0476-1

    • Google Scholar
75. 1. Sanchez G
    2. Dury AY
    3. Murray LM
    4. Biondi O
    5. Tadesse H
    6. El Fatimy R
    7. Kothary R
    8. Charbonnier F
    9. Khandjian EW
    10. Côté J

    (2013) A novel function for the survival motoneuron protein as a translational regulator

    Human Molecular Genetics 22:668–684.

    https://doi.org/10.1093/hmg/dds474

    • Google Scholar
76. 1. Schwartz JC
    2. Wang X
    3. Podell ER
    4. Cech TR

    (2013) RNA seeds higher-order assembly of FUS protein

    Cell Reports 5:918–925.

    https://doi.org/10.1016/j.celrep.2013.11.017

    • Google Scholar
77. 1. Selenko P
    2. Sprangers R
    3. Stier G
    4. Buhler D
    5. Fischer U
    6. Sattler M

    (2001) SMN tudor domain structure and its interaction with the Sm proteins

    Nature Structural Biology 8:27–31.

    https://doi.org/10.1038/83014

    • Google Scholar
78. 1. Smith E
    2. Lin C
    3. Shilatifard A

    (2011) The super elongation complex (SEC) and MLL in development and disease

    Genes & Development 25:661–672.

    https://doi.org/10.1101/gad.2015411

    • Google Scholar
79. 1. Sonenberg N
    2. Hinnebusch AG

    (2009) Regulation of translation initiation in eukaryotes: mechanisms and biological targets

    Cell 136:731–745.

    https://doi.org/10.1016/j.cell.2009.01.042

    • Google Scholar
80. 1. Strahl BD
    2. Briggs SD
    3. Brame CJ
    4. Caldwell JA
    5. Koh SS
    6. Ma H
    7. Cook RG
    8. Shabanowitz J
    9. Hunt DF
    10. Stallcup MR
    11. Allis CD

    (2001) Methylation of histone H4 at arginine 3 occurs in vivo and is mediated by the nuclear receptor coactivator PRMT1

    Current Biology 11:996–1000.

    https://doi.org/10.1016/S0960-9822(01)00294-9

    • Google Scholar
81. 1. Swiercz R
    2. Person MD
    3. Bedford MT

    (2004)

    Ribosomal protein S2 is a substrate for mammalian protein arginine methyltransferase 3 (PRMT3)

    The Biochemical Journal 386:85–91.

    • Google Scholar
82. 1. Takahama K
    2. Kino K
    3. Arai S
    4. Kurokawa R
    5. Oyoshi T

    (2011) Identification of Ewing's sarcoma protein as a G-quadruplex DNA- and RNA-binding protein

    FEBS Journal 278:988–998.

    https://doi.org/10.1111/j.1742-4658.2011.08020.x

    • Google Scholar
83. 1. Takahama K
    2. Takada A
    3. Tada S
    4. Shimizu M
    5. Sayama K
    6. Kurokawa R
    7. Oyoshi T

    (2013) Regulation of telomere length by G-quadruplex telomere DNA- and TERRA-binding protein TLS/FUS

    Chemistry & Biology 20:341–350.

    https://doi.org/10.1016/j.chembiol.2013.02.013

    • Google Scholar
84. 1. Thandapani P
    2. O'Connor TR
    3. Bailey TL
    4. Richard S

    (2013) Defining the RGG/RG motif

    Molecular Cell 50:613–623.

    https://doi.org/10.1016/j.molcel.2013.05.021

    • Google Scholar
85. 1. Wang H
    2. Huang Z-Q
    3. Xia L
    4. Feng Q
    5. Erdjument-Bromage H
    6. Strahl BD
    7. Briggs SD
    8. Allis CD
    9. Wong J
    10. Tempst P
    11. Zhang Y

    (2001) Methylation of histone H4 at arginine 3 facilitates transcriptional activation by nuclear hormone receptor

    Science 293:853–857.

    https://doi.org/10.1126/science.1060781

    • Google Scholar
86. 1. Wolfe AL
    2. Singh K
    3. Zhong Y
    4. Drewe P
    5. Rajasekhar VK
    6. Sanghvi VR
    7. Mavrakis KJ
    8. Jiang M
    9. Roderick JE
    10. Van der Meulen J
    11. Schatz JH
    12. Rodrigo CM
    13. Zhao C
    14. Rondou P
    15. de Stanchina E
    16. Teruya-Feldstein J
    17. Kelliher MA
    18. Speleman F
    19. Porco JA Jr
    20. Pelletier J
    21. Rätsch G
    22. Wendel HG

    (2014) RNA G-quadruplexes cause eIF4A-dependent oncogene translation in cancer

    Nature 513:65–70.

    https://doi.org/10.1038/nature13485

    • Google Scholar
87. 1. Xu J
    2. Wang AH
    3. Oses-Prieto J
    4. Makhijani K
    5. Katsuno Y
    6. Pei M
    7. Yan L
    8. Zheng YG
    9. Burlingame A
    10. Brückner K
    11. Derynck R

    (2013) Arginine methylation initiates BMP-induced smad signaling

    Molecular Cell 51:5–19.

    https://doi.org/10.1016/j.molcel.2013.05.004

    • Google Scholar
88. 1. Yu Z
    2. Chen T
    3. Hébert J
    4. Li E
    5. Richard S

    (2009) A mouse PRMT1 null allele defines an essential role for arginine methylation in genome maintenance and cell proliferation

    Molecular and Cellular Biology 29:2982–2996.

    https://doi.org/10.1128/MCB.00042-09

    • Google Scholar

Thank you for sending your work entitled “Aven recognition of RNA G-quadruplexes regulates translation of the Mixed Lineage Leukemia protooncogenes” for consideration at eLife. Your article has been overall favorably evaluated by James Manley (Senior Editor), a Reviewing Editor, and two reviewers. One of the two reviewers, Ernesto Guccione, has agreed to share his identity.

The Reviewing Editor and the two reviewers, whose comments are below, discussed their reviews before reaching this decision, and the Reviewing Editor has assembled the following comments to help you prepare a revised submission.

The manuscript submitted by Thandapani et al. investigates the mechanism through which the protein Aven inhibits apoptosis leading to the promotion of malignancy. The authors present data supporting the role of Aven as an RNA binding protein through its N-terminal RGG/RG domain and suggest that it binds G-quadruplex RNA structures. Arginine methylation of the RGG/RG domain by PRMT1 was shown to affect RNA binding, association with polyribosomes, and interactions with two Tudor domain proteins (SMN and TDRD3). The authors identify G-quadruplex motifs in two mRNAs (MLL1 and MLL4) as targets of Aven. It also appears that the RGG/RG domain of Aven recruits an RNA helicase that can unwind G-quadruplexes (DHX36) to polyribosomes, and this increases the translation of mRNAs. Taken together the data suggest a model where Aven binds the G-quadruplex motifs in the MLL mRNAs, through its RGG/RG domain. It associates with polysomes dependent on methylation of its RGG/RG domain by PRMT1, and its binding to polysome-associated Tudor domain proteins. Polysome-associated Aven recruits a GQ-unwinding helicase that then promotes translation of MLL1 and MLL4 by destabilizing the G-quadruplexes in their coding sequences. Increased expression of these histone methyltransferases affects transcription of HOX mRNAs and this is associated with the increased cell proliferation and survival function postulated for Aven.

1) The most important issue with this manuscript has to do with incomplete or unclear quantitation of data, use of statistical tests that are not described at all, and lack of demonstration of reproducibility. For example, it needs to be indicated for every experiment how many times it has been performed independently. Blots should be quantified when possible. Error bars are almost never defined (an exception is Figure 5G) and statistical tests are not described. There is no description of “n” numbers or how “n” is defined. It's difficult to evaluate the experiments presented without this information.

2) The paper is not very well referenced. There are missing references that should be cited and in some cases better papers could be cited. The most important of the missing references is Melzer et al. (Cell Death and Differentiation, 19:1435, 2012) in which they find cathepsin D cleavage sites and map the anti-apoptotic function of Aven to the C-terminus once released by proteolysis from the inhibitory N-terminus. I'd like the authors to address this work and discuss their findings in relation it.

3) The authors rely heavily on deletion of the ΔRGG region in Aven. It would be nice to have at least one experiment performed with an R-K mutant instead of the deletion construct.

Reviewer #1:

Figure 1 suggests that endogenous Aven can be methylated by endogenous PRMT1 on several arginines in the N-terminal RGG/RG domain, though a few definitive controls as mentioned below are missing. However, even without all the controls when all the data is considered together it's likely that the authors are correct.

However, relevant to panel (A), it's not correct to say that Aven has no other known domains (in the subsection “Aven RGG/RG motif is methylated by PRMT1”). The authors should cite, or defend why they aren't citing, several publications: one by Esmaili et al. (Cell Cycle, 9:3913, 2010) suggesting a CRM1-dependent NES, one by Hawley et al. (Open Biol. J., 5:6, 2012) finding a BH3 motif (binding to Bcl-X_L), and one from Melzer et al. (Cell Death and Differentiation, 19:1435, 2012) who find cathepsin D cleavage sites and importantly map the anti-apoptotic function of Aven to the C-terminus once released by proteolysis from the inhibitory N-terminus? The authors should comment on these findings.

Figure 1C shows that transfected Myc-Aven can capture endogenous PRMT1 in HEK cells. The figure, legend and text (in the subsection “Aven RGG/RG motif is methylated by PRMT1”) all need modification to indicate that odd numbered lanes lack Myc-Aven.

Figure 1D uses OHT to knockout PRMT1 from a floxed allele in MEFs. The authors suggest that this shows that Myc-Aven is no longer methylated after loss of PRMT1 but a control is missing (mock transfection) which would show the methylated band in lane 1 is actually Myc-Aven and not a contaminating methylated PRMT1 substrate. This still would not control for a co-IPing methylated PRMT1 substrate dependent on the IP of Aven – transfection with Myc-Aven lacking the RGG/RG box would be definitive. Finally, MEFs lacking floxed PRMT1 would be an appropriate negative control to show that OHT alone doesn't inhibit methylation of Aven, i.e. show dependence on the PRMT1.

Figure 1E: Endogenous Aven was IP'ed after tamoxifen-induced deletion of PRMT1, or not. Aven is shown to no longer be phosphorylated if PRMT1 is lost. Again it's possible that OHT treatment alone might lead to decreased Aven phosphorylation independent of PRMT1. Dependence on PRMT1 could be shown using MEFs lacking floxed PRMT1 as a negative control.

Figure 1–figure supplement 1: A line should be added to the text of the paper pointing out the conservation of methylated arginines across species.

Figure 2: the general conclusion is probably correct, that the Aven RGG/RG domain has the potential to bind G-quadruplex forming sequences in vitro even when it's unmethylated. Whether methylation changes the affinity to any degree can't be determined.

The RNA binding assays used in Figure 2 are not described in enough detail. For example, what was the volume of cell lysate to which 400-500 pmol of RNA was added (needed to calculate molarity of the RNA)?

Figure 2D: The statement can't be made that methylated and unmethylated Aven bind equally well to sc1 from the experiment presented. Salt sensitivity is not the same as a Kd. At 150 mM salt either the protein or the RNA needs to be titrated and % bound at each concentration determined to calculate a Kd for methylated vs. unmethylated.

Figure 2E: In this experiment a Kd could be calculated with the right assay since both peptides and RNA concentrations are known. The curves shown do not reach saturation so a solid conclusion about differing affinity can't be made. The method needs to be described including the sequences of the 4 RG motif peptides. RNA should be described as a concentration, not an amount.

Figure 2–figure supplement 1: Though quantitation and evidence of reproducibility are lacking, it does appear that deletion of the RGG box doesn't markedly affect phosphorylation of ATM or CHK2, or it's subcellular localization (though this is not a shuttling assay), or it's dimerization, though this is a double overexpression assay. A small note – the figure legend lists a different anti-pCHK2 antibody than in the figure labeling.

Figure 3 addresses whether Aven interacts with 3 Tudor domain proteins and whether this is dependent on the RGG box and its methylation status. In Figure 3A, double overexpression GST pull-downs suggest the RGG box is required for TDRD3 and SMN. This is extended in (B) and (C) to the capture of endogenous TDRD3 and SMN and finally in (D) co-IP of endogenous Aven with endogenous TDRD3 and SMN is shown, and found to be dependent on OHT treatment. The authors suggest this means it's dependent on methylation of Aven by PRMT1 but dependence on PRMT1 is not shown. Further support for this would be OHT treated WT MEFS, or a second method of knocking down PRMT1, like siPRMT1 treatment. There is nothing in these studies to rule out complex formation via a third protein(s) or RNA that links Aven to the other two proteins, for example, if they are all polysome-associated, and intact polysomes are being IPed. RNAse treatment would be appropriate here. Note that sc1 is not a substitute for the RNAs that Aven might be interacting with in a cell.

Figure 4: The authors demonstrate that a percentage of cellular Aven is associated with polysomes that are actively translating based on the puromycin run-off experiment shown. However, the data that the RGG box affects polysome association is not convincing. If the blots in F and G were exposed a little more will they look like E? How many times was the experiment done? Why does rpS6 disappear completely with puromycin treatment? It should appear in the fractions with the monosome peak. Quantitation of the loss from polysomes with error bars would be appropriate to make the point here. In (K-M) the same concerns apply. As presented I don't find the decrease in polysome-associated Flag-Aven convincing.

Figure 5–figure supplement 1 should be referred to as table supplement 2 or vice versa. I am not clear about the reasoning behind or the significance of the statement that the mRNA encoding domains rich in R's and G's may have propensity to form G-quadruplexes. Nor why that particular consensus was used to query ORFs for potential G-quadruplex forming sequences. Is there literature to support that that query predicts that RNA fold? The rationale should be made clear. I'm particularly worried about the inclusion of C's without restraints in the N1-7 intervals. A stretch of C's will almost certainly form a duplex with nearby runs of G's interrupting the potential for GQ formation. A folding program could supplement this analysis, as could conservation across species that would predict maintenance of the GQ fold. Compensatory mutation to preserve folding would be even stronger evidence. If it's correct that the C-terminus of Aven holds the anti-apoptotic function once released from the N-terminus, it's not clear why the authors are proposing the connection between PG4 binding in mRNAs encoding proteins involved in leukemic cell survival. Could the authors please comment on the findings in Melzer et al. (Cell Death and Differentiation, 19:1435, 2012)?

Figure 5: Again, this is not a proper RNA binding assay (Figure 5C, D). The finding that Aven-ΔRGG no longer binds RNA is not evidence for a specific interaction (Figure 5E, F). The positive charge of the RGG box will lead to nonspecific RNA association under certain conditions, again it depends on the Kd and that data is not available.

Figure 5G: In the CLIP experiments presented in this panel, there is no evidence for a direct interaction. A non-crosslinked control is missing. In addition, the “CLIP” technique as described was done under conditions that the same reference (Huppertz et al.) concluded caused substantial cell toxicity, and no washes of the IP are described. It should not be referred to as “CLIP” in the current form, since there is no attempt to take advantage of the covalent bond introduced by crosslinking which would include rigorous purification of Aven:RNA complexes including stringent washes and SDS-PAGE. Moreover, this is PAR-CLIP, not iCLIP as referenced. The reference should be changed to the methodology developed by the Tuschl lab. An unbiased map of where Aven is bound to RNA in vivo could be obtained by performing the whole CLIP protocol including high throughput sequencing of the RNA tags. However, overexpression of Aven with a Flag tag is not optimal, as RBPs tend to associate with targets of lesser affinity and specificity when overexpressed beyond normal stoichiometry with their RNA targets. The antibody against Aven could be used to perform CLIP on endogenous Aven with unbiased and more comprehensive results.

Figure 7: Panel (B) shows polysome profiles for WT and Aven-null cultured cells. The conclusion is made that the Aven-null cells have less polysomes, but it appears that they have less total ribosomes as well since the monosome peak is decreased. If equal cell numbers were used for the polysome analysis this might suggest a decrease in the total machinery available for protein synthesis. Panels (C)-(F) show the results for only part of the gradient. It would be valuable to see the whole gradient. Moreover, mRNA per fraction can only be plotted as a percentage of the total RNA if the total is determined for the whole gradient. For example, in (C) a large increase in MLL1 mRNA in the WT (blue line) in fractions 1-4 could cause the blue line in fractions 13-14 to actually be less than the red line, which changes the conclusions. If the authors mean that they quantified the mRNA in an aliquot of the lysate applied to the gradient, this should be clarified. For less confusion, real values for each fraction should be shown rather than percentage, which makes the assumption that the total mRNA levels in the presence or absence of Aven are unchanged. It would be nice to see this data again as it supports Figures 6C and 6D.

Figure 8: Again, please include reproducibility (n) and statistical tests and whether the results are statistically robust.

Figure 9: Could the authors comment on whether the small global decrease in initiation in the presence of Flag-Aven is reproducible? (Blue line, loss of polysomes with a gain in the subunits and monosomes not seen for the green and red lines). In (D), where does the DHX36 go if not on polysomes? There doesn't seem to be any difference in total levels in panel (A). Does the blot need to be exposed longer? Fractions 1-4, maybe; it would be nice to see this. It does look like DHX36 has normal polysome association in (E) but I'm again unconvinced that there is a significant difference between the results in panels (E) and (F). Quantitation? Reproducible?

Reviewer #2:

1) Add molecular markers also on the right panel in Figure 1B.

2) could be moved to supplementary figures.

3) One of the key figures in my opinion is Figure 7, where the authors are able to show that the MLL1 and MLL4 are selectively depleted from the heavy polysome fractions. This is performed in the presence or absence of Aven. It would be great to see the same experiment done in siPRMT1 and/or siSMN+siTDR3 conditions. An alternative suggestion would be to move this figure at the end of the manuscript and add also siDHX36.

4) The data presented in Figure 10 do not add much and could be removed (or improved). Questions that come to mind are: is the drug selectively killing cancer cells that overexpress MLL1/MLL4? Or does it kill any cell? What is the toxicity of this drug? How direct is the effect we see on MLL1/MLL4 protein abundance, as opposed to a general toxicity in cell culture?

[Editors’ note: this article was subsequently rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]

Thank you for sending the revised version of your work entitled “Aven recognition of RNA G-quadruplexes regulates translation of the Mixed Lineage Leukemia protooncogenes” for consideration at eLife. Your revised submission has been evaluated by James Manley (Senior Editor), a Reviewing Editor, and one peer reviewer, and the decision was reached after discussions between the Reviewing Editor and the reviewer. Based on our discussions and the individual review below, we regret to inform you that your work cannot be considered further for publication in eLife.

While the editor and the reviewer felt that you had improved many aspects of the manuscript that had been identified as problematic (some new data and clarification of the number of independent replicates and explanation of statistical tests), they also felt that your work still does not sufficiently connect your observations to the biology of the system. The editor and the referee felt that, while many of the details of the function of Aven were defined, the key question as to the biological significance of the RGG box (and its potential interaction with RNA and the downstream consequences) was not adequately addressed. In light of these deficiencies, it was agreed that your manuscript fails to meet the level of significance required for publication in eLife.

Reviewer #1:

In this resubmission the authors have improved the paper considerably with new data and clarification of the number of independent replicates and explanation of statistical tests. Overall, they have addressed most of my experimental concerns.

However, there are two bigger issues that are still bothering me. The background to the paper is that the authors previously found that Aven harbors an N-terminal RGG motif conserved across species. It strikes me that the most important issue, before investigating its function, should be to show that the RGG box is required for the anti-apoptotic (or other) biological functions of Aven in affecting cancer progression. Once established, it then becomes of greater interest to understand the molecular functions of the RGG box. My reading of the paper is that those assays haven't been done and in terms of the ones that have been performed, deletion of the RGG box had no effect. Moreover, as I pointed out in my first review, Melzer et al. have published that the anti-apoptotic function of Aven resides in the C-terminus and the N-terminus (containing the RGG box) acts to repress this activity until cleavage of Aven by cathepsin D releases the active C-terminus. I think this at odds with the idea that the RGG box is central to the anti-apoptotic activity unless I'm missing something. I asked the authors to discuss this and they didn't answer the question. They do add a sentence or two to the paper (in addition to agreeing to mention the Melzer et al. work) in which they say they can barely detect the product of cathepsin D cleavage, therefore it's unlikely to be important to what they are studying, but that doesn't address the issue of demonstrating that the N-terminal RGG box is actually required for the anti-apoptotic properties of Aven.

I think they have skipped the step, or are avoiding the issue, of establishing that the RGG box is important. Because they bioinformatically found an RGG box in Aven, and some RGG boxes bind RNA, and some are methylated by PRMT1, and some have been reported to bind GQ's in vitro, and some GQ's have been identified in RNA, and in some cases may affect translation, the authors have built a story suggesting that this is true for Aven and is relevant for understanding its anti-apoptotic function. They do have good data for a lot of this now, except for the importance of the RGG box in the functions of Aven that make it interesting.

If they had established that the RGG box was critical, this would still be a higher impact paper if they had then used PAR-CLIP or a similar assay to 1) determine if Aven binds RNA in vivo, and 2) generate an unbiased genome wide map of where it binds. The binding sites would not only reveal mRNA targets but would show GQ motif consensus if that's really the in vivo binding site for Aven. The polysome-association and PRMT1 methylation studies then follow to determine the outcome of Aven binding those sites in those mRNAs (which may or may not be translational control) as well as what regulates it. Instead, the paper reads like a very biased expedition to explore a few reported properties of RGG boxes and RNA G-quadruplexes based on prior literature, not all of which is high quality and many conclusions are from in vitro studies with unclear in vivo relevance. I realize we previously discussed this and agreed that it would be unreasonable to ask them to do these experiments (though it does bother me that they have already done half the PAR-CLIP experiment but didn't finish it, which could transform their paper (or prove their assertions correct) fairly easily).

The work is appealing because the authors present a compelling argument that understanding how Aven inhibits apoptosis in cancer cells, and consequently allows their progression when it's overexpressed, has potential for increasing therapeutic avenues for some human cancers. But I'm worried about their reticence to discuss the Melzer paper in regard to the issue of the RGG box being required for the anti-apoptotic function. Given the weak experimental approach (which we agreed not to ask them to change) I don't think it’s unreasonable to at least ask them to fully address evidence in support or in conflict with the idea that the RGG box has a relevant biologic role, to validate their experiments that speak to its function.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your work entitled “Aven recognition of RNA G-quadruplexes regulates translation of the Mixed Lineage Leukemia protooncogenes” for further consideration at eLife. Your article and letter of appeal have been evaluated by James Manley (Senior Editor), a Reviewing Editor, and two reviewers. One of the two reviewers (Reviewer 1) reviewed the earlier versions of your work, while the other (Reviewer 3) has not reviewed this study before and provides a fresh perspective.

Although the ultimate biological relevance of the RGG box, or even translation regulation by Aven, is not yet clear, we are willing to move forwards with acceptance, providing you can address the minor comments of Reviewer 3. We also ask you to provide a final response to Reviewer 1's comments and address them where you can within the Discussion. We are not asking for additional experiments.

Reviewer #1 (abridged):

In their appeal Dr. Richard and co-authors discuss two of the points I raised when re-reviewing the manuscript.

The most important issue is with the second point I raised. To paraphrase my previous summary of that issue, in order to support the importance of this work investigating the function of the Aven RGG box, it should be demonstrated that the RGG box is required for the anti-apoptotic (or other) biological functions of Aven in affecting cancer progression. Once that's established, it then becomes of greater interest to understand the molecular functions of the RGG box. In general those assays haven't been done and in terms of the ones that have been performed, deletion of the RGG box had no effect. Moreover, as I pointed out in my first review, Melzer et al. have published that the anti-apoptotic function of Aven resides in the C-terminus and the N-terminus (containing the RGG box) acts to repress this activity until cleavage of Aven by cathepsin D releases the active C-terminus. This is at odds with the idea that the RGG box is central to the anti-apoptotic activity.

In their appeal, they have provided a greater response to that but are still missing/avoiding the point and lacking in the validation experiments that could support the importance of the function of the RGG box.

The authors cite literature that many groups agree that full-length Aven has anti-apoptotic activity. However, my concern isn't so much whether full length Aven has an anti-apoptotic role but the other finding of Melzer et al.: that robust anti-apoptotic activity can be mapped to fragment of Aven lacking the RGG box, which would be at odds with the current paper finding that the RGG box is mediating the anti-apoptotic activity. What I wanted the Richard group to discuss is the Melzer et al. finding that the RGG-box containing N-terminus of Aven mediates homodimerization, and that its release from FL Aven by cathepson D cleavage activates the C-terminus as an anti-apoptotic molecule. Whether or not the N-terminus is really inhibitory, or FL Aven has anti-apoptotic activity, the Richard group should provide some insight as to why Melzer et al. find that the RGG deleted form of Aven has anti-apoptotic activity.

The authors now include a new figure for the reviewer (Author response image 1) using an assay for Aven function which is clonogenic survival of U2OS cells, purporting to show that the RGG box is required. The assay shows that without endogenous Aven, the cells are killed with increasing UV irradiation. Overexpression of Aven protects them. There is a small decrease in the ability of Aven to do so if the RGG box is deleted but it is only significant at the highest UV dose and is a modest effect, perhaps 2-fold. Panel B shows a Western blot of the expression of Aven and Aven-ΔRGG and it is quite clear that there's much more WT Aven produced than Aven-ΔRGG. It is impossible to know if the effects they see are due to Aven dose or to the RGG box. CRISPR-mediated deletion of the RGG box from endogenous Aven would be more compelling. They've used that technology to make Aven null cells and perhaps could use the same technique to make an elegant RGG null form of Aven.

The authors also cite three figures in their paper (Figures 3, 7, and 10) that they claim support the role of the RGG box, however, Figure 3 only shows the RGG box is needed for protein-protein association. Figure 10 shows that the ability of Aven to regulate translation of a luciferase reporter containing a G-quadruplex forming sequence is dependent on the RGG box. Nicely shown, but very in vitro and not an assay of cell survival. Figure 7 is quoted by the authors as relevant, but I think they mean Figure 5 G-H which are the RIP and partly finished PAR-CLIP experiment to show the RGG box is important for association with MLL1 and MLL4 RNA.

As I wrote previously, I think the Richard group has “skipped the step, or are avoiding the issue, of establishing that the RGG box is important […]. They do have good data for a lot of this now, except for the importance of the RGG box in the functions of Aven that make it interesting.”

Reviewer #3:

This article concerns the role of Aven in regulating the translation of leukemia-relevant mRNAs. Aven is an RGG box protein that promotes cell survival and is linked to osteosarcoma and AML. The strength of the manuscript is the use of multiple lines of evidence to justify the majority of the conclusions. The weakness is the qualitative nature of many of the assays. Nevertheless, all of the data presented are internally consistent, and I am left with the impression that the data as presented do largely support the authors' conclusions. Both mechanistic and biological insights arise from the study, which increases overall interest to a broad readership. While the manuscript would be improved by a more thorough study of Aven-associated mRNAs through a true PAR-CLIP study, or by further studies into the biological relevance of the handful of newly identified regulated mRNAs presented here, I agree with the authors' rebuttal that addition of such data to this manuscript would increase the scope well beyond what should be found in a single paper.

Ultimately, it is interesting that Aven binds to G-quartets, that its association with polysomes is methylation dependent, and that it regulates the translation of at least a few leukemia specific mRNAs, ultimately leading to reduced amounts of HOX gene transcription. And while a broader study would enhance the impact, it is hard to imagine filling this manuscript full of even more experiments.

Minor comments:

1) It seems disingenuous to call the photocrosslinking IP experiments presented in this paper “PAR-CLIP”. The use of photoactivatable nucleotides to crosslink proteins to RNA to assist in recovery by IP preceded the paper by Hafner et al. by many years. PAR-CLIP, as published, is touted as a transcriptome-wide approach.

2) The description of the RGG box as an “RNA Recognition Motif” (in the Introduction) could be misleading, as the RRM is a specific RNA-binding domain that is not the same as the RGG box. Though many proteins contain both.

3) The rationale behind the selection of SMN and TDRD3 for further study was not clear to this reviewer. Have either been implicated in osteosarcoma or myeloid leukemia? There are 71 Tudor domain proteins in humans according to EMBL SMART database.

4) Can sc1 RNA bind to Aven if it hasn't been induced to fold into a G-quartet?

5) What fraction of the proteins associated with Aven in the SILAC experiments are RNA-dependent?

6) I am surprised that real-time PCR was used to detect enrichment of the G-quadruplex structures in the photocrosslinked IPs. I would have guessed that the G quadruplex would have prevented efficient reverse transcription. The authors should comment on this point.

7) The significant digits in Figure 2E should be re-evaluated.

[Editors’ note: the author responses to the first round of peer review follow.]

1) The most important issue with this manuscript has to do with incomplete or unclear quantitation of data, use of statistical tests that are not described at all, and lack of demonstration of reproducibility. For example, it needs to be indicated for every experiment how many times it has been performed independently. Blots should be quantified when possible. Error bars are almost never defined (an exception is Figure 5G) and statistical tests are not described. There is no description of “n” numbers or how “n” is defined. It's difficult to evaluate the experiments presented without this information.

We now provide additional quantification and statistical tests, and indicate the number of independent experiments performed.

2) The paper is not very well referenced. There are missing references that should be cited and in some cases better papers could be cited. The most important of the missing references is Melzer et al. (Cell Death and Differentiation 19:1435 (2012)) in which they find cathepsin D cleavage sites and map the anti-apoptotic function of Aven to the C-terminus once released by proteolysis from the inhibitory N-terminus. I'd like the authors to address this work and discuss their findings in relation it.

We have added additional functional regions in the diagram of Aven protein in Figure 1A including a putative BH3 domain identified using bioinformatics and a functional nuclear export signal (NES). We also cite additional pertinent papers, such as Hawley et al., 2012; Melzer et al., 2012 and Esmaili et al., 2010.

Melzer et al. (2012) describe that in addition to the full length Aven (MW∼50kDa), MCF-7 and HEK293T cells express a truncated Aven C-terminal protein (MW∼30kDa) that is thought to be generated by Cathepsin D cleavage. It also appears that this Aven C-terminal fragment has potent anti-apoptotic activity, as judged by overexpression studies. We used the same ProScience antibody as Melzer et al. (2012) which recognizes the Aven C-terminus. An ∼30kDa Aven C-terminal protein was notably absent in MOLT-4 (Figure 6A) and CCRF-CEM (Figure 6B) cells. In HEK293T cells, however, after prolonged exposure, we do see a very minor band (∼34kDa) that is absent in the Aven CRISPR/CAS null cells that could represent cleaved C-terminal Aven fragment described by Melzer et al.

To reflect these findings, we have added the following text in the Results section: “A minor Aven ∼34kDa fragment (Figure 7A, denoted by asterisk) was observed in HEK293T cells and likely represents a cathepsin D cleaved fragment reported previously [ #8127]”. We have also added the following to the Discussion: “It was reported that an N-terminal deleted fragment of Aven cleaved by cathepsin D harbors […]. Thus cathepsin D mediated cleavage of Aven is unlikely involved in the regulation of translational control described herein.”

3) The authors rely heavily on deletion of the ΔRGG region in Aven. It would be nice to have at least one experiment performed with an R-K mutant instead of the deletion construct.

We have generated Aven R-K mutant and present data in Figure 10 which shows that this mutant closely mimics effects observed in experiments using Aven RGG.

Reviewer #1:

Figure 1 suggests that endogenous Aven can be methylated by endogenous PRMT1 on several arginines in the N-terminal RGG/RG domain, though a few definitive controls as mentioned below are missing. However, even without all the controls when all the data is considered together it's likely that the authors are correct.

However, relevant to panel (A), it's not correct to say that Aven has no other known domains (in the subsection “Aven RGG/RG motif is methylated by PRMT1”). The authors should cite, or defend why they aren't citing, several publications: one by Esmaili et al. (Cell Cycle, 9:3913, 2010) suggesting a CRM1-dependent NES, one by Hawley et al. (Open Biol. J., 5:6, 2012) finding a BH3 motif (binding to Bcl-X_L), and one from Melzer et al. (Cell Death and Differentiation, 19:1435, 2012) who find cathepsin D cleavage sites and importantly map the anti-apoptotic function of Aven to the C-terminus once released by proteolysis from the inhibitory N-terminus? The authors should comment on these findings.

Please see response to 1) above.

Figure 1C shows that transfected Myc-Aven can capture endogenous PRMT1 in HEK cells. The figure, legend and text (subsection “Aven RGG/RG motif is methylated by PRMT1”, second paragraph) all need modification to indicate that odd numbered lanes lack Myc-Aven.

The legend for Figure 1C and the Results section were edited, as suggested.

The Results section was also modified to include the following: “PRMT1 was present in anti-Myc immunoprecipitates of the Myc-Aven transfected cells, but not in the empty vector transfected cells (Figure 1C, lanes 3, 4).”

Figure 1D uses OHT to knockout PRMT1 from a floxed allele in MEFs. The authors suggest that this shows that Myc-Aven is no longer methylated after loss of PRMT1 but a control is missing (mock transfection) which would show the methylated band in lane 1 is actually Myc-Aven and not a contaminating methylated PRMT1 substrate. This still would not control for a co-IPing methylated PRMT1 substrate dependent on the IP of Aven – transfection with Myc-Aven lacking the RGG/RG box would be definitive. Finally, MEFs lacking floxed PRMT1 would be an appropriate negative control to show that OHT alone doesn't inhibit methylation of Aven, i.e. show dependence on the PRMT1.

Myc-Aven migrates ∼70kDa, as it was engineered to harbor 5 Myc-tags to distinguish it from endogenous Aven that normally migrates as ∼50kDa. Thus, we are confident of the identity of the Myc-Aven band, however, we have added an anti-Myc immunoblot (Figure 1E; panel below ASYM25) of the immunoprecipitated proteins to demonstrate that the band at ∼70kDa is indeed Myc-Aven.

Figure 1E: Endogenous Aven was IP'ed after tamoxifen induced deletion of PRMT1, or not. Aven is shown to no longer be phosphorylated if PRMT1 is lost. Again it's possible that OHT treatment alone might lead to decreased Aven phosphorylation independent of PRMT1. Dependence on PRMT1 could be shown using MEFs lacking floxed PRMT1 as a negative control.

To address the question related to potential inadvertent effects caused by the use of OHT, we now add an alternative approach to show hypomethylation of Aven. HEK293T cells were transfected with Myc-Aven and depleted of PRMT1 using siRNAs (new Figure 1D). We obtain similar data as using floxed PRMT1 MEFs treated with OHT i.e. that Myc-Aven is recognized by ASYM25b in PRMT1-proficient, but not PRMT1-deficient cells. These data demonstrate that Aven is indeed a substrate of PRMT1 in vivo.

Figure 1–figure supplement 1: A line should be added to the text of the paper pointing out the conservation of methylated arginines across species.

In Figure 1–figure supplement 1 we have boxed the conserved arginines. The following sentences were added to the Results section: “Arginines R5, R8, R11, R37, R50 and R63 are also conserved in murines (Figure 1–figure supplement 1). Taken together, these findings demonstrate that the Aven RGG/RG motif is methylated by PRMT1 on conserved arginines.”

Figure 2: The general conclusion is probably correct, that the Aven RGG/RG domain has the potential to bind G-quadruplex forming sequences in vitro even when it's unmethylated. Whether methylation changes the affinity to any degree can't be determined.

The RNA binding assays used in Figure 2 are not described in enough detail. For example, what was the volume of cell lysate to which 400-500 pmol of RNA was added (needed to calculate molarity of the RNA)?

Figure 2D: The statement can't be made that methylated and unmethylated Aven bind equally well to sc1 from the experiment presented. Salt sensitivity is not the same as a Kd. At 150 mM salt either the protein or the RNA needs to be titrated and % bound at each concentration determined to calculate a Kd for methylated vs. unmethylated.

Figure 2E: In this experiment a Kd could be calculated with the right assay since both peptides and RNA concentrations are known. The curves shown do not reach saturation so a solid conclusion about differing affinity can't be made. The method needs to be described including the sequences of the 4 RG motif peptides. RNA should be described as a concentration, not an amount.

To address these issues, we performed a new set of experiments where we obtain a Kd for two separate Aven RGG/RG peptides (see Figure 2E and Figure 2–figure supplement 1 panel D). Arginine methylation did not affect the binding affinities, as the Aven TriRG peptides (Kd ∼80-90nM; Figure 2E) and the DiRGG peptides (Kd∼173-200nM; Figure 2–figure supplement 1 panel D) did not have different Kds.

Figure 2–figure supplement 1: Though quantitation and evidence of reproducibility are lacking, it does appear that deletion of the RGG box doesn't markedly affect phosphorylation of ATM or CHK2, or it's subcellular localization (though this is not a shuttling assay), or it's dimerization, though this is a double overexpression assay. Small note – the figure legend lists a different anti-pCHK2 antibody than in the figure labeling.

The activation of ATM and CHK2 and the cytoplasmic localization was performed n = 3. We have clearly stated the number of independent replicates that were performed throughout the revised version of the manuscript, including the legend of Figure 2–figure supplement 1. We have corrected the pCHK2 antibody to be pT68.

Figure 3 addresses whether Aven interacts with 3 Tudor domain proteins and whether this is dependent on the RGG box and its methylation status. In Figure 3A, double overexpression GST pull-downs suggest the RGG box is required for TDRD3 and SMN. This is extended in (B) and (C) to the capture of endogenous TDRD3 and SMN and finally in (D) co-IP of endogenous Aven with endogenous TDRD3 and SMN is shown, and found to be dependent on OHT treatment. The authors suggest this means it's dependent on methylation of Aven by PRMT1 but dependence on PRMT1 is not shown. Further support for this would be OHT treated WT MEFS, or a second method of knocking down PRMT1, like siPRMT1 treatment. There is nothing in these studies to rule out complex formation via a third protein(s) or RNA that links Aven to the other two proteins, for example, if they are all polysome-associated, and intact polysomes are being IPed. RNAse treatment would be appropriate here. Note that sc1 is not a substitute for the RNAs that Aven might be interacting with in a cell.

We agree that there may be a third protein or RNA in the complex. To clarify these issues we added the following sentence to the Discussion: “The Aven interaction with TDRD3 and SMN may require other protein or RNA components in the complex for enhanced association.”

As suggested by the reviewer, we have performed a new experiment (Figure 3D), showing that siPRMT1 disrupts the interaction between FLAG-Aven and endogenous SMN and TDRD3.

Figure 4: The authors demonstrate that a percentage of cellular Aven is associated with polysomes that are actively translating based on the puromycin run-off experiment shown. However, the data that the RGG box affects polysome association is not convincing. If the blots in F and G were exposed a little more will they look like E? How many times was the experiment done? Why does rpS6 disappear completely with puromycin treatment? It should appear in the fractions with the monosome peak. Quantitation of the loss from polysomes with error bars would be appropriate to make the point here. In (K-M) the same concerns apply. As presented I don't find the decrease in polysome-associated Flag-Aven convincing.

To address the reviewer’s concerns, we now present both short and long exposures of the immunoblots for FLAG-Aven and FLAG-Aven RGG (Figure 4) and have modified the text accordingly to addresses exposure time and reproducibility. We also show quantification of these data by densitometry (Figure 4–figure supplement 2). We have also edited the legend for Figure 4.

Figure 5–figure supplement 1 should be referred to as table supplement 2 or vice versa. I am not clear about the reasoning behind or the significance of the statement that the mRNA encoding domains rich in R's and G's may have propensity to form G-quadruplexes. Nor why that particular consensus was used to query ORFs for potential G-quadruplex forming sequences. Is there literature to support that that query predicts that RNA fold? The rationale should be made clear. I'm particularly worried about the inclusion of C's without restraints in the N1-7 intervals. A stretch of C's will almost certainly form a duplex with nearby runs of G's interrupting the potential for GQ formation. A folding program could supplement this analysis, as could conservation across species that would predict maintenance of the GQ fold. Compensatory mutation to preserve folding would be even stronger evidence. If it's correct that the C-terminus of Aven holds the anti-apoptotic function once released from the N-terminus, it's not clear why the authors are proposing the connection between PG4 binding in mRNAs encoding proteins involved in leukemic cell survival. Could the authors please comment on the findings in Melzer et al. (Cell Death and Differentiation, 19:1435, 2012)?

We now refer to the table as Figure 5–figure supplement 1. We have edited our statements to clarify why we searched for PG4: “Next, we reasoned that since RGG/RG motifs are encoded by G-rich sequences […] [fx1]and it revealed ∼1600 PG4 in human ORFs (Figure 5–figure supplement 1).”

We calculated cG/cC score and the RNAfold using v2.1.7 and these data are added in (Figure 5–figure supplement 1). To reflect this, the following text was added to the Results section: “We also provide a cG/cC score […] predicting G4 formation than the cG/cC ratio [Beaudoin, 2014 #8043].”

Figure 5: Again, this is not a proper RNA binding assay (Figure 5C, D). The finding that Aven-ΔRGG no longer binds RNA is not evidence for a specific interaction (Figure 5E, F). The positive charge of the RGG box will lead to nonspecific RNA association under certain conditions, again it depends on the Kd and that data is not available.

We have modified the text accordingly (“We examined whether Aven associated with the PG4 RNA […] in Figure 2E with a fluoresceinated PG4 (sc1) to be in the 80-200nM range).

Figure 5G: In the CLIP experiments presented in this panel, there is no evidence for a direct interaction. A non-crosslinked control is missing. In addition, the “CLIP” technique as described was done under conditions that the same reference (Huppertz et al.) concluded caused substantial cell toxicity, and no washes of the IP are described. It should not be referred to as “CLIP” in the current form, since there is no attempt to take advantage of the covalent bond introduced by crosslinking which would include rigorous purification of Aven:RNA complexes including stringent washes and SDS-PAGE. Moreover, this is PAR-CLIP, not iCLIP as referenced. The reference should be changed to the methodology developed by the Tuschl lab. An unbiased map of where Aven is bound to RNA in vivo could be obtained by performing the whole CLIP protocol including high throughput sequencing of the RNA tags. However, overexpression of Aven with a Flag tag is not optimal, as RBPs tend to associate with targets of lesser affinity and specificity when overexpressed beyond normal stoichiometry with their RNA targets. The antibody against Aven could be used to perform CLIP on endogenous Aven with unbiased and more comprehensive results.

We have edited the Methods section to indicate that we performed PAR-CLIP, not iCLIP. Indeed, our cells were healthy after the 8 h 4SU treatment. We have also edited the reference (Hafner et al., 2010 rather than ; Huppertz et al., 2014) and the total of 4 washes were performed, which is stated in the revised Methods section.

Moreover, as suggested we have performed an additional PAR-CLIP experiment with or without UV crosslinking and the IPs were performed with rabbit control IgG or anti-Aven (Proscience) antibodies. We observe binding of Aven to PG4 sequences and only when UV is utilized (Figure 5H).

Figure 7: panel (B) shows polysome profiles for WT and Aven-null cultured cells. The conclusion is made that the Aven-null cells have less polysomes, but it appears that they have less total ribosomes as well since the monosome peak is decreased. If equal cell numbers were used for the polysome analysis this might suggest a decrease in the total machinery available for protein synthesis. Panels (C)-(F) show the results for only part of the gradient. It would be valuable to see the whole gradient. Moreover, mRNA per fraction can only be plotted as a percentage of the total RNA if the total is determined for the whole gradient. For example, in (C) a large increase in MLL1 mRNA in the WT (blue line) in fractions 1-4 could cause the blue line in fractions 13-14 to actually be less than the red line, which changes the conclusions. If the authors mean that they quantified the mRNA in an aliquot of the lysate applied to the gradient, this should be clarified. For less confusion, real values for each fraction should be shown rather than percentage, which makes the assumption that the total mRNA levels in the presence or absence of Aven are unchanged. It would be nice to see this data again as it supports Figures 6C and 6D.

We used an identical number of cells and loaded the same amount of material (i.e. 10 ODs at 254nm) on the gradients. We then integrated the area under curves and normalized curves based on the observed differences in the area, which is a standard method of normalization of polysome absorbance profiles (Gandin et al., JOVE, 2014). Aven null cells exhibited comparable absorbance profiles as Aven-proficient cells (Figure 7B). To accommodate the reviewer’s request, the absorbance profiles for the entire gradient are now presented throughout the manuscript. Also, we observe a clear shift of the MLL1 and MLL4 mRNAs, but not β-actin into the lighter polysomal fractions in cells lacking Aven as compared to Aven-proficient cells (Figure 7C-E). We also include new data showing that siPRMT1 and siDHX36 induces similar shifts in MLL1 and MLL4 mRNAs towards lighter polysomal fractions, but exert only a marginal effect on polysomal loading of β-actin mRNA, as compared to control cells (Figure 8C-E; 9D-F). To analyze qPCR data we used relative standard curve method as described in ABI User Bulletin # 4.

Figure 8: Again, please include reproducibility (n) and statistical tests and whether the results are statistically robust.

The following information is now provided in the legend of Figure 10 (was Figure 8 previously), panels B and C.

“Error bars represent standard deviation values. The experiments were performed three independent times (n = 3) and each independent experiment was performed in technical triplicates. The significance was measured by ANOVA followed by post hoc comparison using Tukey test. *p < 0.05.” (Legend of Figure 10B)

“Error bars represent standard deviation values. The experiments were performed three independent times (n = 3) and each independent experiment was performed in technical triplicates. The significance was measured by ANOVA followed by post hoc comparison using Tukey test. *p < 0.05, n.s. non-significant.” (Legend of Figure 10C)

Figure 9: Could the authors comment on whether the small global decrease in initiation in the presence of Flag-Aven is reproducible? (Blue line, loss of polysomes with a gain in the subunits and monosomes not seen for the green and red lines). In (D), where does the DHX36 go if not on polysomes? There doesn't seem to be any difference in total levels in panel (A). Does the blot need to be exposed longer? Fractions 1-4, maybe; it would be nice to see this. It does look like DHX36 has normal polysome association in (E) but I'm again unconvinced that there is a significant difference between the results in panels (E) and (F). Quantitation? Reproducible?

Combining the comments from both reviewers, we have performed additional polysomal fractionations and we monitored the levels of MLL1 and MLL4 mRNAs in each fraction. We observe a striking difference in cells depleted of DHX36 (new Figure 9). This combined with the luciferase assay builds a strong case that DHX36 stimulates translation of MLL1 and MLL4 mRNAs.

Reviewer #2:

1) Add molecular markers also on the right panel in Figure 1B.

Molecular mass markers were added to the right panel as well.

2) Figure 4 Panel C and D could be moved to supplementary figures.

Panels C and D were moved to Figure 4–figure supplement 1.

3) One of the key figures in my opinion is Figure 7, where the authors are able to show that the MLL1 and MLL4 are selectively depleted from the heavy polysome fractions. This is performed in the presence or absence of Aven. It would be great to see the same experiment done in siPRMT1 and/or siSMN+siTDR3 conditions. An alternative suggestion would be to move this figure at the end of the manuscript and add also siDHX36.

As suggested by the reviewer, we have reorganized our figures and we now show that MLL1 and MLL4 are selectively depleted from heavy polysomes fractions with siPRMT1 (new Figure 8) and siDHX36 (new Figure 9). We also repeated the polysomal fractionation with siSMN and siTDRD3 (Figure 4H).

4) The data presented in Figure 10 do not add much and could be removed (or improved). Questions that come to mind are: is the drug selectively killing cancer cells that overexpress MLL1/MLL4? Or does it kill any cell? What is the toxicity of this drug? How direct is the effect we see on MLL1/MLL4 protein abundance, as opposed to a general toxicity in cell culture?

Figure 10 was removed, as we agree with the reviewer that the toxicity of the G4 ligand drug may be due to the presence of many other G4 motifs present in the genome.

[Editors' note: the author responses to the second round of peer review follow.]

We would like to appeal to the rejection of our manuscript entitled “Aven recognition of RNA Gquadruplexes regulates translation of the Mixed Lineage Leukemia protooncogenes ”. Reviewer #2 found our manuscript suitable for presentation, whereas Reviewer #1 found that we were largely responsive to her/his comments, and that the revised version of the manuscript was vastly improved. Reviewer#1 felt, however, that our revised manuscript was not suitable for publication in eLife, which she/he rationalized by two major points:

I) “I realize we previously discussed this and agreed that it would be unreasonable to ask them to do these experiments (though it does bother me that they have already done half the PAR-CLIP experiment but didn't finish it, which could transform their paper (or prove their assertions correct) fairly easily). […] Given the weak experimental approach (which we agreed not to ask them to change) I don't think it’s unreasonable to at least ask them to fully address evidence in support or in conflict with the idea that the RGG box has an a relevant biologic role to validate their experiments that speak to its function.”

We trust that a genome-wide PAR-CLIP experiment is out of scope of our present manuscript as we demonstrate that the RGG box is involved in G-quartet binding using well-established biochemical assays (Figures 2 and 5) and show that Aven RGG box is required for recruitment of G-quartet containing MLL1, MLL4 and reporter mRNAs to the polysomes and translation as evidenced by simultaneously monitoring expression of corresponding proteins (Figures 4, 7, 10). Moreover, it appears that during the first round of reviews consensus was reached that a genome-wide PAR-CLIP experiment was indeed out of scope of our study.

“I realize we previously discussed this and agreed that it would be unreasonable to ask them to do these experiments”.

Therefore, we were unaware of these issues as they were not brought up in the review and thus we feel that is extremely unfair to bring these issues up in the second round of reviews and use it as a major basis to reject the paper.

In addition, Reviewer#1 states that “PAR-CLIP experiment but didn't finish it, which could transform their paper (or prove their assertions correct) fairly easily)”.

Genome-wide PAR-CLIP experiments are cumbersome and require extensive validation (Riley & Steitz, Mol Cell, 49:601-604, 2013). For this reason, we feel that a standalone study is merited to carefully carry out genome-wide PAR-CLIP analysis, and validate and functionally characterize resulting hits (including all possible scenarios such as that some mRNAs may be bound, but not translationally regulated by Aven), rather than just to provide a list of potential Aven mRNA targets in our present study. Therefore, based on the data presented in our present study, we are undertaking genome-wide PAR-CLIP experiments from total and polysomal extracts, wherein in parallel we are performing ribosome profiling study to identify not just mRNAs that are AVEN bound, but also those whose polysome-association and translation are regulated by AVEN. As we hope you will appreciate, this approach is not widely used and thus we need to take significant time and be prudent not to produce yet another list of false positives. We thus anticipate that this study will take several years to complete.

II) Moreover, the reviewer was worried about apparent discrepancies between our Melzer et al., paper, which highlighted the role of N-terminal truncated Aven as the major anti-apoptotic form of the protein.

In contrast to Melzer et al., and consistent with our data, full-length Aven was shown to exhibit anti-apoptotic effects (Chau et al., Mol Cell, 2000; Kutuk et al., Eur J. Cancer, 2010; Eibmann/Mezler et al.,Oncogene, 2012; Ouzounova et al., BMC Genomics, 2013). Furthermore, Melzer et al. (Eibmann/Mezler et al., Oncogene, 2012) published another paper where overexpression of full-length Aven driven by the lck promoter opposes apoptosis in primary thymocytes. Accordingly, we did not observe the truncated form of Aven in MOLT-4 and CCRF-CEM cells (Figure 6). Taken together, these findings clearly indicate that Aven plays multiple mutually not excluding, and likely cell type specific, roles in promoting cell proliferation and survival. Indeed, we present a hitherto unprecedented role of Aven in promoting proliferation and long-term survival of leukemic cells (Aven is overexpressed in the vast majority of leukemia) by bolstering the translation of MLL1 and MLL4 which as we demonstrate using multiple experimental approaches is dependent on the integrity of the RGG motif (Figures 3, 7, 10).

Moreover, we performed clonogenic assay that shows that RGG motif is also required for clonogenic survival of U2OS cells (Author response image 1). We also modified the Discussion to highlight the discrepancies between Meltzer et al. and our study.

Collectively, we feel that we provide a plethora of evidence for the requirement of the RGG box for AVENs biological function and provide an underpinning mechanism. We also feel that Reviewer #1’s concerns need a significant period of time to complete and merit being addressed in a separate study. For the reasons stated above, we feel that the decision to reject our manuscript should be reconsidered.

Author response image 1

Download asset Open asset

[Editors’ note: the author responses to the re-review follow.]

Reviewer #1 (abridged):

[…] As I wrote previously, I think the Richard group has “skipped the step, or are avoiding the issue, of establishing that the RGG box is important […]. They do have good data for a lot of this now, except for the importance of the RGG box in the functions of Aven that make it interesting.”

The following text in the Discussion addresses the RGG box and the Melzer paper: “It was reported that an N-terminal […] including those encoding MLL1 and MLL4 required for cell survival.”

Reviewer #3:

1) It seems disingenuous to call the photocrosslinking IP experiments presented in this paper “PAR-CLIP”. The use of photoactivatable nucleotides to crosslink proteins to RNA to assist in recovery by IP preceded the paper by Hafner et al. by many years. PAR-CLIP, as published, is touted as a transcriptome-wide approach.

We have replaced PAR-CLP with photocrosslinking IP experiments. We also cite the original paper Cahill et al., 2000 (Steitz), rather than the Hafner paper.

2) The description of the RGG box as an “RNA Recognition Motif” (in the Introduction) could be misleading, as the RRM is a specific RNA-binding domain that is not the same as the RGG box. Though many proteins contain both.

We removed “RNA Recognition Motif” and edited the sentence to read: “The RGG/RG motif, also called RGG box, was shown to bind RNA…”.

3) The rationale behind the selection of SMN and TDRD3 for further study was not clear to this reviewer. Have either been implicated in osteosarcoma or myeloid leukemia? There are 71 Tudor domain proteins in humans according to EMBL SMART database.

We added the following to the Results section: “Although there are many Tudor domain containing proteins, methylated RGG/RG motifs are known to interact specifically with the Tudor domains of TDRD3, SMN and SPF30 (Selenko et al., 2001; Cote and Richard, 2005).”

4) Can sc1 RNA bind to Aven if it hasn't been induced to fold into a G-quartet?

This was not tested.

5) What fraction of the proteins associated with Aven in the SILAC experiments are RNA-dependent?

We did not perform a RNase control for this experiment. However, we added the following statement in the Results section: “Of the 146 proteins enriched with Aven, but not AvenΔRGG, ∼23% were ribosomal proteins and ∼10% were RNA binding proteins, their association is likely RNA-dependent (Figure 4–figure supplement 1).”

6) I am surprised that real-time PCR was used to detect enrichment of the G-quadruplex structures in the photocrosslinked IPs. I would have guessed that the G quadruplex would have prevented efficient reverse transcription. The authors should comment on this point.

The fold induction observed between Aven and the G4 sequences are likely under represented due to the fact that the crosslinks impede the reverse transcription reactions (Huppertz et al., 2014)

7) The significant digits in Figure 2E should be re-evaluated.

The significant digits were verified, as suggested.

Author details

1. Palaniraja Thandapani

   1. Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada
   2. Bloomfield Center for Research on Aging, Lady Davis Institute for Medical Research, McGill University, Montréal, Canada
   3. Department of Oncology, McGill University, Montréal, Canada
   4. Department of Medicine, McGill University, Montréal, Canada

   Contribution

   PT, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Contributed equally with

   Jingwen Song

   Competing interests

   The authors declare that no competing interests exist.

2. Jingwen Song

   1. Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada
   2. Bloomfield Center for Research on Aging, Lady Davis Institute for Medical Research, McGill University, Montréal, Canada
   3. Department of Oncology, McGill University, Montréal, Canada
   4. Department of Medicine, McGill University, Montréal, Canada

   Contribution

   JS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Contributed equally with

   Palaniraja Thandapani

   Competing interests

   The authors declare that no competing interests exist.

3. Valentina Gandin

   Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada

   Contribution

   VG, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

4. Yutian Cai

   1. Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada
   2. Department of Biochemistry, McGill University, Montréal, Canada

   Contribution

   YC, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

5. Samuel G Rouleau

   Département de Biochimie, Université de Sherbrooke, Sherbrooke, Canada

   Contribution

   SGR, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

6. Jean-Michel Garant

   Département de Biochimie, Université de Sherbrooke, Sherbrooke, Canada

   Contribution

   J-MG, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

7. Francois-Michel Boisvert

   Département d'Anatomie et de Biologie Cellulaire, Faculté de Médecine et des Sciences de la Santé, Pavillon de Recherche Appliquée au Cancer, Université de Sherbrooke, Sherbrooke, Canada

   Contribution

   F-MB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

8. Zhenbao Yu

   1. Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada
   2. Bloomfield Center for Research on Aging, Lady Davis Institute for Medical Research, McGill University, Montréal, Canada
   3. Department of Oncology, McGill University, Montréal, Canada
   4. Department of Medicine, McGill University, Montréal, Canada

   Contribution

   ZY, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

9. Jean-Pierre Perreault

   Département de Biochimie, Université de Sherbrooke, Sherbrooke, Canada

   Contribution

   J-PP, Conception and design, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

10. Ivan Topisirovic

    1. Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada
    2. Department of Biochemistry, McGill University, Montréal, Canada
    3. Department of Oncology, McGill University, Montréal, Canada

    Contribution

    IT, Conception and design, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

11. Stéphane Richard

    1. Terry Fox Molecular Oncology Group, Segal Cancer Center, Jewish General Hospital, Montréal, Canada
    2. Bloomfield Center for Research on Aging, Lady Davis Institute for Medical Research, McGill University, Montréal, Canada
    3. Department of Oncology, McGill University, Montréal, Canada
    4. Department of Medicine, McGill University, Montréal, Canada

    Contribution

    SR, Conception and design, Analysis and interpretation of data, Drafting or revising the article

    For correspondence

    stephane.richard@mcgill.ca

    Competing interests

    The authors declare that no competing interests exist.

• 3,533

  Page views

• 790

  Downloads

• 65

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.