1. Cell Biology
  2. Chromosomes and Gene Expression

Vasohibin1, a new mouse cardiomyocyte IRES trans-acting factor that regulates translation in early hypoxia

  1. Fransky Hantelys
  2. Anne-Claire Godet
  3. Florian David
  4. Florence Tatin
  5. Edith Renaud-Gabardos
  6. Françoise Pujol
  7. Leila H Diallo
  8. Isabelle Ader
  9. Laetitia Ligat
  10. Anthony K Henras
  11. Yasufumi Sato
  12. Angelo Parini
  13. Eric Lacazette
  14. Barbara Garmy-Susini
  15. Anne-Catherine Prats  Is a corresponding author
  1. UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, France
  2. UMR 1031-STROMALAB, Inserm, CNRS ERL5311, Etablissement Français du Sang-Occitanie (EFS), National Veterinary School of Toulouse (ENVT), Université de Toulouse, UPS, France
  3. UMR 1037-CRCT, Inserm, CNRS, Université de Toulouse, UPS, Pôle Technologique-Plateau Protéomique, France
  4. UMR 5099-LBME, CBI, CNRS, Université de Toulouse, UPS, France
  5. Tohoku University, Japan
https://doi.org/10.7554/eLife.50094
Share this article
Cite this article
  1. Fransky Hantelys
  2. Anne-Claire Godet
  3. Florian David
  4. Florence Tatin
  5. Edith Renaud-Gabardos
  6. Françoise Pujol
  7. Leila H Diallo
  8. Isabelle Ader
  9. Laetitia Ligat
  10. Anthony K Henras
  11. Yasufumi Sato
  12. Angelo Parini
  13. Eric Lacazette
  14. Barbara Garmy-Susini
  15. Anne-Catherine Prats
(2019)
Vasohibin1, a new mouse cardiomyocyte IRES trans-acting factor that regulates translation in early hypoxia
eLife 8:e50094.
https://doi.org/10.7554/eLife.50094
  2. Download BibTeX
  3. Download .RIS

Abstract

Hypoxia, a major inducer of angiogenesis, triggers major changes in gene expression at the transcriptional level. Furthermore, under hypoxia, global protein synthesis is blocked while internal ribosome entry sites (IRES) allow specific mRNAs to be translated. Here, we report the transcriptome and translatome signatures of (lymph)angiogenic genes in hypoxic HL-1 mouse cardiomyocytes: most genes are induced at the translatome level, including all IRES-containing mRNAs. Our data reveal activation of (lymph)angiogenic factor mRNA IRESs in early hypoxia. We identify vasohibin1 (VASH1) as an IRES trans-acting factor (ITAF) that is able to bind RNA and to activate the FGF1 IRES in hypoxia, but which tends to inhibit several IRESs in normoxia. VASH1 depletion has a wide impact on the translatome of (lymph)angiogenesis genes, suggesting that this protein can regulate translation positively or negatively in early hypoxia. Translational control thus appears as a pivotal process triggering new vessel formation in ischemic heart.

Introduction

Hypoxia constitutes a major stress in different pathologies including both cancer and ischemic pathologies where artery occlusion leads to hypoxic conditions. In all of these pathologies, hypoxia induces a cell response that stimulates angiogenesis to re-feed starved cells with oxygen and nutrients (Pouysségur et al., 2006). Recently it has been shown that lymphangiogenesis is also induced by hypoxia (Morfoisse et al., 2014). Hypoxia-induced (lymph)angiogenesis is mediated by strong modification of gene expression at both transcriptional and post-transcriptional levels (Pouysségur et al., 2006; Holcik and Sonenberg, 2005). A major mode of regulation of gene expression is mediated at the transcriptional level by the hypoxia inducible factor 1 (HIF1), a transcription factor stabilized by oxygen deprivation, that activates transcription from promoters containing hypoxia-responsive elements (HRE). One of the well-described HIF1 targets is vascular endothelial growth factor A (VEGFA), a major angiogenic factor (Forsythe et al., 1996; Pagès and Pouysségur, 2005). However, two other major angiogenic or lymphangiogenic growth factors, fibroblast growth factor 2 (FGF2) and VEGFC, respectively, are induced by hypoxia in a HIF-independent manner by a translational mechanism, indicating the importance of the post-transcriptional regulation of gene expression in this process (Morfoisse et al., 2014; Conte et al., 2008).

Translational control of gene expression plays a crucial role in the stress response. In particular, translation of mRNAs by the classical cap-dependent mechanism is silenced, whereas alternative translation mechanisms allow enhanced expression of a small group of messengers that are involved in the control of cell survival (Holcik and Sonenberg, 2005; Baird et al., 2006; Spriggs et al., 2008). One of the major alternative mechanisms that is able to overcome this global inhibition of translation by stress depends on internal ribosome entry sites (IRESs), which correspond to RNA structural elements that allow the direct recruitment of the ribosome onto mRNA. As regards the molecular mechanisms of IRES activation by stress, several studies have reported that RNA-binding proteins, called IRES trans-acting factors (ITAFs), are able to stabilize the adequate RNA conformation, thus allowing ribosome recruitment (Faye and Holcik, 1849; Godet et al., 2019; Liberman et al., 2015; Mitchell et al., 2003; Morfoisse et al., 2016). Interestingly, subcellular relocalization of ITAFs plays a critical role in IRES-dependent translation (Lewis and Holcik, 2008). Indeed, many RNA-binding proteins are known to shuttle between nucleus and cytoplasm, and it has been reported that cytoplasmic relocalization of ITAFs such as PTB, PCBP1, RBM4 or nucleolin is critical to activate IRES-dependent translation (Godet et al., 2019; Morfoisse et al., 2016; Lewis and Holcik, 2008). By contrast, other ITAFs such as hnRNPA1, may have a negative impact on IRES activity when accumulating in the cytoplasm (Lewis et al., 2007). However, how ITAFs participate in the regulation of the hypoxic response remains a challenging question.

IRESs are present in the mRNAs of several (lymph)angiogenic growth factors in the FGF and VEGF families, suggesting that the IRES-dependent mechanism might be a major way to activate angiogenesis and lymphangiogenesis during stress (Morfoisse et al., 2014; Godet et al., 2019; Morfoisse et al., 2016; Huez et al., 1998; Martineau et al., 2004; Stein et al., 1998; Vagner et al., 1995). However, most studies of the role of hypoxia in the regulation of gene expression have been performed in tumoral hypoxia, although it has been reported that tumoral angiogenesis leads to the formation of abnormal vessels that are non-functional, differing strongly from non-tumoral angiogenesis that induces formation of functional vessels (Jain, 2005). This suggests that the regulation of gene expression in response to hypoxia may be different in cancer versus ischemic pathologies. In particular, the role of IRESs in the control of gene expression in ischemic heart, the most frequent ischemic pathology, remains to be elucidated.

Here, we analyzed the transcriptome and the translatome of (lymph)angiogenic growth factors in hypoxic cardiomyocytes, and studied the regulation of IRES activities in early and late hypoxia. Data show that in cardiomyocyte, (lymph)angiogenic growth factors are mostly regulated at the translational level. Interestingly, IRESs of several mRNAs in the FGF and VEGF families are activated in early hypoxia in contrast to the IRESs of non-angiogenic messengers. We also looked for ITAFs governing IRES activation in hypoxia, and identified vasohibin1 (VASH1) as a new ITAF that is able to activate the FGF1 IRES in hypoxic cardiomyocytes but not the other IRESs studied here. VASH1 depletion has also a wide impact on the recruitment of (lymph)angiogenic factor mRNAs into polysomes, suggesting that this protein can regulate translation positively or negatively in early hypoxia.

Results

Most (lymph)angiogenic genes are not induced at the transcriptome level of hypoxic cardiomyocytes

In order to analyze the expression of angiogenic and lymphangiogenic growth factors in hypoxic cardiomyocytes, the HL-1 cell line was chosen: although immortalized, it keeps the beating phenotype specific to cardiomyocyte (Claycomb et al., 1998). HL-1 cells were submitted to increasing durations of hypoxia, from 5 min to 24 hr, and their transcriptome was analyzed on a Fluidigm Deltagene PCR array targeting 96 genes of angiogenesis, lymphangiogenesis and/or stress (Figure 1, Figure 1—figure supplement 1, Supplementary file 1). Data showed a significant increase of Vegfa, PAI-1 and apelin (Apln) mRNA levels, with a peak at 8 hr of hypoxia for Vegfa and PAI1 and 24 hr for Apln. These three genes are well-described HIF1 targets (Forsythe et al., 1996; Kietzmann et al., 1999; Ronkainen et al., 2007). However, although 56% of the detected genes were induced over 1.5-fold, few of them were strongly induced. Furthermore, the mRNA levels of several major angio- or lymphangiogenic factors, such as FGF2 and VEGFC, were strongly decreased after 4 hr or 8 hr of hypoxia. These data indicate that the transcriptional response to hypoxia in cardiomyocytes is not the major mechanism controlling the expression of (lymph)angiogenic factors, suggesting that post-transcriptional mechanisms are involved.

Figure 1 with 1 supplement see all
Download asset Open asset
(Lymph)angiogenic genes are not drastically induced at the transcriptome level in hypoxic cardiomyocytes.

Total RNA was purified from HL-1 cardiomyocytes submitted to increasing durations (from 5 min to 24 hr) of hypoxia at 1% O2, as well as from normoxic cardiomyocytes as a control. cDNAs were synthesized and used for a Fluidigm deltagene PCR array dedicated to genes related to (lymph)angiogenesis or stress (Supplementary file 6). Analysis was performed in three biological replicates (cell culture well and cDNA), each of them measured in three technical replicates (PCR reactions). Relative quantification (RQ) of gene expression during hypoxia was calculated using the 2–ΔΔCT method with normalization to 18S rRNA and to normoxia. mRNA levels are presented as histograms for the times of 4 hr, 8 hr and 24 hr, as the fold change of repression (red) or induction (green) normalized to normoxia. Non-regulated mRNAs are represented in blue. The threshold for induction was set at 1.5. When the RQ value is inferior to 1, the fold change is expressed as −1/RQ. The percentage of repressed, induced, and non-regulated mRNAs is indicated for each duration of hypoxia. For shorter durations of 5 min to 2 hr, the percentages are shown in Figure 1—figure supplement 1. The detailed values for all of the durations of hypoxia are presented in Supplementary file 1.

mRNAs of most (lymph)angiogenic genes are recruited into polysomes in hypoxic cardiomyocytes

On the basis of the fact that mRNAs that are present in polysomes are actively translated, we tested the hypothesis of translational induction by analyzing the recruitment of mRNAs into polysomes. This experiment was performed in early and late hypoxia. The polysome profile showed that translational activity in normoxic HL-1 cells was low but decreased after 4 hr of hypoxia, with a shift of the polysome to monosome ratio from 1.55 to 1.40 (Figure 2A). Eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1) appeared as a single band and its phosphorylation profile did not change upon hypoxia, suggesting that this protein is already hypophosphorylated in normoxia in these cells (Figure 2—figure supplement 1A and B). By contrast, translation blockade was confirmed by the strong phosphorylation of eIF2α (Figure 2B, Figure 2—figure supplement 1C). 94% of the genes of the (lymph)angiogenic array showed a more sustained recruitment into polysomes under hypoxic conditions (Figure 2C, Supplementary file 2). This translational induction not only targets genes that encode major angiogenic factors and their receptors (e.g. Vegfa, Fgf1, Pdgfa, Fgfr3, Vegfr2), but also genes involved in cardiomyocyte survival in ischemic heart (Igf1, Igf1r) or in inflammation (BAI1, Tgfb). These data suggest that, in cardiomyocytes, the main response of (lymph)angiogenic genes to early hypoxia is not transcriptional, but translational.

Figure 2 with 1 supplement see all
Download asset Open asset
mRNAs of most (lymph)angiogenic genes are mobilized into polysomes in hypoxic cardiomyocytes.

(A-C) In order to isolate translated mRNAs, polysomes from HL-1 cardiomyocytes in normoxia or after 4 hr of hypoxia at 1% O2 were purified on a sucrose gradient, as described in 'Materials and methods'. Analysis was performed using a fluidigm PCR array from three biological replicates (cell culture well and cDNA), each of them measured in three technical replicates (PCR reactions). P/M ratio (polysome/monosome) was determined by delimiting the 80S and polysome peaks by taking the lowest plateau values between each peak and by calculating the area under the curve (AUC). Then the sum of area values of the four polysome peaks was divided by the area of the 80S peak (A). Translation blockade was measured by eIF2α phosphorylation quantification by Jess capillary Simple Western, normalized to the Jess quantification of total proteins (as described in 'Materials and methods'). Three independent experiments were done; a representative experiment is shown (B). RNA was purified from polysome fractions and from cell lysate before loading. cDNA and PCR arrays were performed as in Figure 1. Polysomes profiles are presented for normoxic and hypoxic cardiomyocytes. Relative quantification (RQ) of gene expression during hypoxia was calculated using the 2–ΔΔCT method with normalization to 18S rRNA and to normoxia. mRNA levels (polysomal RNA/total RNA) are shown as fold change of repression (red) or induction (green) in hypoxia normalized to normoxia as in Figure 1C. The threshold for induction was set at 1.5. When the RQ value is inferior to 1, the fold change is expressed as −1/RQ. The detailed values are available in Supplementary file 2.

IRES-containing mRNAs are more efficiently mobilized into polysomes under hypoxic conditions

IRES-dependent translation has been reported to drive the translation of several mRNAs in stress conditions (Morfoisse et al., 2014; Holcik and Sonenberg, 2005; Conte et al., 2008; Morfoisse et al., 2015). Thus, we focused on the regulation of the different IRES-containing mRNAs present in the Fluidigm array (Figure 3). Interestingly, the only IRES-containing mRNA to be significantly induced by hypoxia at the transcriptome level was Vegfa (Figure 3A and Figure 3—figure supplement 1). Expression of the apelin receptor gene (Aplnr), which is presumably devoid of IRES but transcriptionally induced during hypoxia, is also shown for comparison.

Figure 3 with 1 supplement see all
Download asset Open asset
IRES-containing mRNAs are more efficiently associated with polysomes in hypoxic conditions.

(A, B) RQ values for the IRES-containing mRNA transcriptome (A) and translatome (B) extracted from the PCR arrays shown in Figures 1 and 2. The gene Aplnr (apelin receptor) was chosen as a control without an IRES. Vegfc and Fgf2 mRNAs, which are repressed in the transcriptome, were below the detection threshold in polysomes (ND). Histograms correspond to means ± standard deviation, with values for hypoxia compared to those for normoxia by a bilateral Student's test *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

The polysome recruitment of these IRES-containing mRNAs is shown in Figure 3B. Clearly, Fgf1, Vegfa, Vegfd, Cyr61, Hif1a and Igf1r mRNAs were recruited into polysomes under hypoxia, accumulating to levels 2 to 3 times those in normoxia, suggesting an important induction in terms of translation. By contrast, the recruitment of Aplnr mRNA into polysomes decreased about three times. The data are not available for Fgf2 and Vegfc mRNAs, which were not detectable. These results indicate that hypoxia in cardiomyocytes, although blocking global cap-dependent translation, induces the translation of all detectable IRES-containing angiogenic factor mRNAs. This mechanism occurs as soon as 4 hr after oxygen deprivation, thus corresponding to an early event in the hypoxic response.

IRESs of (lymph)angiogenic factor mRNAs are activated during early hypoxia

To confirm that the polysome recruitment of IRES-containing mRNAs actually corresponds to a stimulation of IRES-dependent translation, IRESs from FGF and VEGF mRNAs were introduced into a bicistronic dual luciferase gene expression cassette (Figure 4A). Two IRESs from non-angiogenic mRNAs, c-myc and EMCV IRESs, were used as controls. A negative control without IRES was provided by a hairpin inserted between the two cistrons (Créancier et al., 2000). The well-established bicistronic vector strategy, previously validated by us and others, allows the measurement of IRES activity, which is revealed by expression of the second cistron, LucF (Morfoisse et al., 2014; Créancier et al., 2000). The bicistronic cassettes were subcloned into lentivectors because HL-1 cells are not efficiently transfected by plasmids but can be easily transduced by lentivectors, with an efficiency of more than 80% (not shown). HL-1 cardiomyocytes were first transduced with the lentivector containing the FGF1 IRES and the kinetics of IRES-dependent translation and protein expression were than examined after between 1 hr and 24 hr of hypoxia. Luciferase activities were measured from cell extracts and IRES activities were reported as the LucF/LucR luminescence ratio.

Data showed an increase in IRES activity between 4 hr and 8 hr of hypoxia whereas this activity decreased between 16 hr and 24 hr of hypoxia (Figure 4B). The expression of endogenous FGF1 was analyzed after 8 hr of hypoxia. FGF1 protein quantification (normalized to total proteins) showed that IRES induction correlates with an increased expression of FGF1 protein (Figure 4C). This is also consistent with the increase of FGF1 mRNA recruitment into polysomes observed above (Figure 3B, Supplementary file 2). To determine whether this transient induction could affect other IRESs, HL-1 cells were then transduced by the complete series of lentivectors described above (Figure 4A) and submitted to 4, 8 or 24 hr of hypoxia. Results showed an increase in all FGF and VEGF IRES activities in early hypoxia, except for VEGFA IRESa (4 hr and/or 8 hr), whereas the c-myc and EMCV IRESs were activated only in late hypoxia after 24 hr (Figure 4D). By contrast, VEGFA and VEGFD IRES activities decreased after 24 hr of hypoxia. The hairpin control was not induced (Supplementary file 3J). These data revealed two groups of IRESs: a first group including IRESs from (lymph)angiogenic growth factor mRNAs (except for VEGFA IRESa) that are activated during early hypoxia, and a second group including ‘non-angiogenic’ c-myc and EMCV IRESs, which are activated in late hypoxia.

Figure 4
Download asset Open asset
IRESs from (lymph)angiogenic factor mRNAs are activated in early hypoxia.

(A–D) To measure IRES-dependent translation during hypoxia, HL-1 cardiomyocytes were transduced with bicistronic dual luciferase lentivectors (termed ‘Lucky-Luke’) containing different IRESs cloned between the genes of renilla (LucR) and firefly (LucF) luciferase (A). In bicistronic vectors, the translation of the first cistron LucR is cap-dependent, whereas translation of the second cistron LucF is IRES-dependent (Créancier et al., 2000). Cardiomyocytes transduced by the CRF1AL+ lentivector Lucky-Luke reporter containing FGF1 IRES were submitted to a hypoxia time-course (0 hr, 1 hr, 2 hr, 4 hr, 6 hr, 8 hr, 16 hr and 24 hr) and data from each time point were compared to those from normoxia with a non-parametric Mann-Whitney test (B). Endogenous FGF1 protein expression was measured by Jess capillary Simple Western of extracts of cardiomyocytes in normoxia or those submitted to 8 hr of hypoxia (normalized to Jess quantification of total proteins as described in 'Materials and methods'). Three independent experiments were performed; a representative experiment is shown (C). HL-1 cardiomyocytes transduced by different Lucky-Luke constructs were submitted to 4 hr, 8 hr or 24 hr of hypoxia and their luciferase activities were measured. IRES activities during hypoxia, expressed as LucF/LucR ratio, are normalized to those during normoxia. Histograms correspond to means ± standard deviation of the mean, data from hypoxic cardiomyocytes compared to those from normoxic cardiomyocytes with a non-parametric Mann-Whitney (M-W) test: *p<0.05, **p<0.01, ***<0.001, ****p<0.0001. For each IRES, the mean was calculated from nine cell culture biological replicates, each of them being the mean of three technical replicates (27 technical replicates in total but the M-W test was performed with n = 9). Detailed values of biological replicates are presented in Supplementary file 3. A no-IRES control was also used and values are presented in Supplementary file 3J.

Identification of IRES-bound proteins in hypoxic cardiomyocytes reveals vasohibin1 as a new RNA-binding protein

Early activation of angiogenic factor IRESs during hypoxia suggested that specific ITAFs may be involved between 4 and 8 hr of hypoxia. In an attempt to identify such ITAFs, we used the biomolecular analysis coupled to mass spectrometry (BIA-MS) technology, which had been validated for ITAF identification in two previous studies (Morfoisse et al., 2016; Ainaoui et al., 2015). Biotinylated RNAs corresponding to FGF1 (early activation), VEGFAa (no activation) and EMCV IRESs (late activation) were used as probes for BIA-MS. Hooked proteins from normoxic and hypoxic HL-1 cells were then recovered and identified (Figure 5A–B, Supplementary file 4). Surprisingly, except for nucleolin bound to VEGFAa and EMCV IRES in normoxia, no known ITAF was identified as being bound to these IRESs in normoxia or in hypoxia. Interestingly, besides several proteins unrelated to (lymph)angiogenesis, we detected the presence of vasohibin1 (VASH1), a protein described as an endothelial cell-produced angiogenesis inhibitor and for its role in stress tolerance and cell survival (Figure 5C) (Sato, 2012; Sato, 2015). This secreted protein has never been reported to have any RNA-binding activity. VASH1 interaction with the FGF1 IRES was detected after 4 hr or 8 hr of hypoxia, but not under normoxia (Supplementary file 4). This protein also interacted with the EMCV IRES both in normoxia and in hypoxia but not with the VEGFA IRES. In order to address the RNA-binding potential of VASH1, we performed an in silico analysis of VASH1 protein sequence that predicted two conserved RNA-binding domains (RBD) in the N- and C-terminal parts of the full-length protein, respectively (Figure 5C, Figure 5—figure supplement 1A and B). The direct interaction of VASH1 with FGF1, VEGFAa and EMCV IRESs was assessed by surface plasmon resonance using the full-length recombinant 44-kDa protein, resulting in the measurement of affinity constants of 6.5 nM, 8.0 nM and 9.6 nM, respectively (Figure 5D–F). These data indicate that VASH1 exhibits a significant RNA-binding activity.

Figure 5 with 1 supplement see all
Download asset Open asset
Identification of IRES-bound proteins in hypoxic cardiomyocytes reveals vasohibin1 as a new RNA-binding protein.

(A–F) Biotinylated IRES RNAs were transcribed in vitro and immobilized on the sensorchip of the BIAcore T200 optical biosensor device (A). Total cell extracts from normoxic or hypoxic HL-1 cardiomyocytes were injected into the device. Complex formation and dissociation were measured (see 'Materials and methods') (B). Bound proteins were recovered as described in 'Materials and methods'. and identified by mass spectrometry (LC-MS/MS) after tryptic digestion. The list of proteins bound in normoxia and hypoxia to FGF1, VEGFAa and EMCV IRESs is shown in Supplementary file 4. VASH1 protein was identified as being bound to FGF1 (hypoxia) and EMCV IRESs (hypoxia and normoxia), but not to VEGFA IRES. A diagram of VASH1 RNA-binding properties is shown, with VASH1 isoforms described by Sonoda et al. (2006) (C). The predicted RNA-binding domains (RBD1 and RBD2) shown in Figure 5‐figure supplement 1 that are conserved in mouse and human (C). Recombinant full-length 44-kDa VASH1 was injected into the Biacore T200 device containing immobilized FGF1 (D), VEGFAa (E) or EMCV (F) IRES as above. The affinity constants (KD) were calculated (D–F) with a single cycle kinetics (SCK) strategy.

Vasohibin1 is translationally induced in early hypoxia and is localized in nuclear and cytoplasmic foci

VASH1 has been previously described for its expression in endothelial cells but never in cardiomyocytes (Sato, 2012). The present BIA-MS study provides evidence that it is expressed in HL-1 cardiomyocytes (Supplementary file 4). We analyzed the regulation of VASH1 expression during hypoxia: Vash1 mRNA level strongly decreases after 4 hr of hypoxia whereas it is slightly upregulated after 8 hr (Supplementary file 1, Figure 6A). By contrast, analysis of Vash1 mRNA recruitment into polysomes showed a strong increase at 4 hr of hypoxia (about 7-fold) (Figure 6B), whereas Vash1 mRNA was not detectable in polysomes after 24 hr of hypoxia (Supplementary file 2). This indicates that Vash1 mRNA translation is strongly induced in early hypoxia. This was confirmed by capillary Simple Western immunodetection, which showed that VASH1 protein expression increases after 4 hr of hypoxia (Figure 6C). Moreover, VASH1 subcellular localization was analyzed by immunocytochemistry: VASH1 appeared as foci in both cytoplasm and nucleus (Figure 6D). The number of foci did not change, but their size significantly increased in hypoxia (Figure 6E and F).

Figure 6
Download asset Open asset
Vasohibin1 is translationally induced in early hypoxia and is localized in nuclear and cytoplasmic foci.

(A–D) VASH1 expression was analyzed in HL-1 cardiomyocytes subjected to hypoxia at the transcriptome and translatome levels. A fluidigm RT qPCR array (Supplementary file 2) was performed with two biological replicates (cell culture and cDNA), each of them measured in two technical replicates (PCR reactions). Detailed values at 4 hr and 24 hr are presented in Supplementary file 2. As for Figure 2, total RNA was purified from the cell lysate of cardiomyocytes in normoxia or submitted to 4 hr, 8 hr or 24 hr of hypoxia (A). Polysomal RNA was purified from cardiomyocytes in normoxia or after 4 hr of hypoxia, from pooled heavy fractions containing polysomes (fractions 19–27) (B). Histograms correspond to mean ± standard deviation of the mean, with two-tailed t-test, *p<0.05, **p<0.01, ***<0.001 used to compare data from hypoxic and normoxic cardiomyocytes. VASH1 protein expression was measured by capillary Simple Western of extracts from cardiomyocytes in normoxia or submitted to 4 hr of hypoxia (C). VASH1 was immunodetected in HL-1 cardiomyocytes in normoxia or after 4 hr of hypoxia (D). DAPI staining allows to detect VASH1 nuclear localization (MERGE). VASH1 foci in the nucleus are shown in purple and those in the cytoplasm in green using Imaris software. The number of VASH1 foci was quantified in the nucleus and in the cytoplasm in normoxia and after 4 hr of hypoxia (n = 4–5 images with a total cell number of 149 in normoxia and 178 in hypoxia) (E). Boxplots of volume of vasohibin foci in normoxia and hypoxia (F). All foci above 0.5 μm3 were counted. Whiskers mark the 10% and the 90% percentiles with the mean in the center. One-way Anova with Tukey’s comparisons test was applied.

Vasohibin1 is a new ITAF that is active in early hypoxia

The putative ITAF function of VASH1 was assessed by a knock-down approach using an siRNA smartpool (siVASH1). Transfection of HL-1 cardiomyocytes with siVASH1 was able to knock-down VASH1 mRNA with an efficiency of 73% (Figure 7A). The knock-down of VASH1 protein measured by capillary Western was only 59% (Figure 7B). This moderate knock-down efficiency was probably due to the long half-life of VASH1, superior to 24 hr (Figure 7—figure supplement 1). The effect of VASH1 knock-down was analyzed in HL-1 cells transduced with different IRES-containing bicistronic lentivectors in normoxia or after 8 hr of hypoxia. In normoxia, VASH1 knock-down generated a moderate increase of activity for several IRESs (13–16%), which was significant for VEGFD and EMCV IRESs (Figure 7C). By contrast, in hypoxia, VASH1 knock-down resulted in a strong decrease of FGF1 IRES activity, by 64%, whereas it did not significantly affect the other IRESs (Figure 7D). These data showed that VASH1 behaves as an activator of FGF IRES in hypoxia, whereas it tends to inhibit several IRESs in normoxia (Figure 7C).

Figure 7 with 1 supplement see all
Download asset Open asset
Vasohibin1 is a new ITAF that is active in early hypoxia.

(A, B) VASH1 knock-down was performed in HL-1 cardiomyocytes using short-interfering (siRNA) smartpools targeting VASH1 (siVASH1) or control (siControl). VASH1 mRNA level was measured by RT-qPCR (A), and VASH1 protein expression was analyzed by the capillary Simple Western method using an anti-VASH1 antibody and quantified by normalization to total proteins. The experiments were reproduced twice, giving identical results. One of the two experiments is shown (B). Knock-down of VASH-1 was performed on cardiomyocytes transduced by a set of IRES-containing lentivectors used in Figure 4. (C, D) After 8 hr of hypoxia, IRES activities (LucF/LucR ratio) were measured in cell extracts from normoxic (C) and hypoxic cardiomyocytes (D). The IRES activity values have been normalized to the control siRNA. Histograms correspond to means ± standard deviation of the mean, and a non-parametric Mann-Whitney test was used to identify significant change from control levels: *p<0.05, **p<0.01. For each IRES the mean was calculated for nine cell culture biological replicates, each of these being the mean of three technical replicates (27 technical replicates in total but the M-W test was performed with n = 9). Detailed values of biological replicates are presented in Supplementary file 5.

Vasohibin1 has a wide impact on the recruitment of (lymph)angiogenesis mRNAs into polysomes

In order to evaluate the possibility of a wider role for VASH1 in translational control, the (lymph)angiogenic factor mRNA polysome profile was analyzed in siVASH1-treated HL-1 cells in normoxia and after 8 hr of hypoxia (Figure 8, Supplementary file 7). VASH1 knock-down strongly affected the mobilization into polysomes of IRES-containing mRNAs, in normoxia and in hypoxia: Vegfd mRNA recruitment increased in normoxia, in concordance with the data shown in Figure 7C. Igf1r mRNA recruitment into polysomes decreased in hypoxia, whereas recruitment of Hif1a and Vegfa mRNAs increased. Unfortunately, Ffg1 and Vegfd mRNAs were not detected in hypoxia, whereas Fgf2 and Vegfc mRNAs were detected neither in normoxia nor in hypoxia in this array experiment, probably because they are poorly expressed. Globally, VASH1 depletion activated the polysome recruitment of 22% and 44% of the detected mRNAs in normoxia and in hypoxia, respectively, whereas it inhibited the recruitment into polysomes of 41% versus 29% of detected mRNAs. Although this approach does not provide information about the mechanism of action, it strongly suggests a wide impact of VASH1, direct or indirect, on translation control. Furthermore, these data confirm that VASH1 has a dual role and can be either an activator or an inhibitor of translation.

Figure 8
Download asset Open asset
VASH1 depletion has both activating and inhibiting effects on mRNA recruitment into polysomes.

HL-1 cardiomyocytes were treated with siVASH1 or siControl and submitted to 8 hr of hypoxia, or maintained in normoxia. RNA was purified from polysome fractions and from cell lysate before loading. cDNA and PCR arrays were performed as in Figure 1. Relative quantification (RQ) of gene expression during hypoxia was calculated using the 2–ΔΔCT method with normalization to 18S rRNA and to SiControl. mRNA levels (polysomal RNA/total RNA) are shown as fold change of repression (red) or induction (green) in siVASH1 cells normalized to SiControl-treated cells. Non-regulated mRNAs are represented in blue. The threshold for induction was set at 1.5. When the RQ value is inferior to 1, the fold change is expressed as −1/RQ. The detailed values are available in Supplementary file 7.

Discussion

The present study highlights the crucial role of translational control in cardiomyocyte response to hypoxia. Up to now, although a few genes had been described for their translational regulation by hypoxia, it was thought that most genes are transcriptionally regulated. Here, we show that translational control, revealed by mRNA recruitment into polysomes during hypoxia, regulates the majority of the genes involved in angiogenesis and lymphangiogenesis. IRES-dependent translation appears to be a key mechanism in this process, as we show that most of the (lymph)angiogenic mRNAs that are known to contain an IRES are upregulated during hypoxia. Furthermore, our data reveal that IRESs of angiogenic factor mRNAs are activated during early hypoxia, whereas the IRESs of non angiogenic mRNAs are activated in late hypoxia. We have identified an angiogenesis- and stress-related protein, VASH1, as a new ITAF that is responsible for the activation of the FGF1 IRES in early hypoxia, whereas it tends to inhibit other IRES activities in normoxia. VASH1 depletion has a positive or negative impact on the recruitment of many (lymph)angiogenesis mRNAs into polysomes, suggesting that this protein is widely involved, directly or not, in translational control in response to stress.

Pathophysiological impact of a moderate stimulation of translation

A striking feature of our data is that the stimulation of IRES activities by hypoxia in cardiomyocytes is moderate, only by 1.3–1.7-fold. Are such small changes in growth factor expression sufficient to alter cellular programs? Several reports demonstrate that the answer is affirmative. An example is VEGFC IRES activation by hypoxia previously shown in tumor cells (Morfoisse et al., 2014). A 2–3-fold increase in endogenous VEGFC expression has a drastic effect on lymphatic vessel growth. In another study, a 2-fold increase in VEGFD IRES activity resulting from heat shock was sufficient to increase lymphatic vessel diameter (Morfoisse et al., 2016). The FGF1 IRES is activated by 1.7–2-fold during myoblast differentiation, and this is sufficient to promote myotube formation controlled by FGF1 (Ainaoui et al., 2015). We have also observed that cellular IRESs have often been reported previously to have a moderate activity in cell culture, whereas they can be much more active than a viral IRES and drastically regulated in vivo (Créancier et al., 2000). This can be explained by the use of cells that are immortalized and not in their physiological environment, which renders them less sensitive to stimuli than cells in vivo. Globally, cellular IRESs show a lower degree of activation than viral IRESs, as illustrated by Braunstein et al. (2007), who report that the HIF1α IRES is stimulated by 1.6-fold during hypoxia, whereas the VEGFA IRES is stimulated by 2-fold and the EMCV IRESs by 3.5-fold.

Translational control in tumoral versus non-tumoral hypoxia

Most studies of gene expression in response to stress have been performed at the transcriptome level in tumoral cells of different origins, whereas the present study is focused on cardiomyocytes. HL-1 cardiomyocytes are immortalized but still exhibit the beating phenotype (Claycomb et al., 1998). Thus, this cell model, although not perfectly mimicking cardiomyocyte behavior in vivo, is still close to a physiological state. The strong translational response to hypoxia revealed by our data, which differs from the transcriptional response usually observed in tumor cells, may reflect mechanisms occurring in cells that are not engaged in the cell transformation process leading to cancer, or at least not too far along this process. Indeed, HL-1 cells respond to hypoxia very early, whereas the various murine or human tumor cell lines described in other reports require a longer period of hypoxia for IRES-dependent translation to be stimulated. In human breast cancer BT474 cells, VEGFA, HIF and EMCV IRESs are all activated after 24 hr of hypoxia (Braunstein et al., 2007). In murine 4T1 and LLC cells (breast and lung tumor, respectively), as well as in human CAPAN-1 pancreatic adenocarcinoma, the VEGFA and VEGFC IRESs are activated after 24 hr of hypoxia whereas the EMCV IRES is not activated (Morfoisse et al., 2014). The same observation of late activation in 4T1 cells has been made for the FGF1 IRES, whereas this IRES is activated in early hypoxia in HL-1 cardiomyocytes (AC Godet and AC Prats, unpublished data) (Figure 4). These observations suggest that many tumoral cell lines that develop resistance to hypoxia are not able to govern subtle the regulations of gene expression in early hypoxia observed in HL-1 cells.

VASH1, an ITAF of early hypoxia

We also consider the hypothesis that the important process of translational regulation observed in our study may be cardiomyocyte-specific. In such a case, IRES-dependent translation would depend on cell-type-specific ITAFs as well as the early response to hypoxia. These results are of great importance in regard to the acute stress response in ischemic heart that is necessary for recovery. By contrast, a delayed chronic response is known to be deleterious for heart healing (Silvestre et al., 2008). In agreement with this hypothesis, VASH1 expression is cell-type-specific: described up to now as endothelial-specific, this protein is not expressed in tumoral cells (Sato, 2012). In the present study, we show that this cell-type specificity extends to cardiomyocytes. Consistent with our data, VASH1 has been described as a key actor in striated muscle angio-adaptation (Kishlyansky et al., 2010). This protein may thus have a role in the early hypoxic response in a limited number of cell types. The ITAF role of VASH1 identified here is physiologically relevant if one considers the function of VASH1 in angiogenesis and stress tolerance (Sato, 2015). According to previous reports, VASH1 is induced during angiogenesis in endothelial cells and halts this process, but its overexpression also renders the same cells resistant to senescence and cell death induced by stress (Sato, 2015). Furthermore, it has been reported that VASH1 is induced after 3 hr of cell stress at the protein level but not at the transcriptional level in endothelial cells (Miyashita et al., 2012). This is in agreement with our observation in cardiomyocytes where VASH1, although downregulated in the transcriptome in early hypoxia, is more efficiently recruited in polysomes at the same time (Figure 6).

It is noteworthy that VASH1 itself seems to be induced translationally by stress (Figure 6) (Miyashita et al., 2012). In endothelial cells, Miyashita et al. (2012) report that the protein HuR upregulates VASH1 by binding to its mRNA. HuR may bind to an AU-rich element present in the 3' untranslated region of the VASH1 mRNA. However, in other studies, HuR has also been described as an ITAF, thus it is possible that VASH1 itself may be induced by an IRES-dependent mechanism (Godet et al., 2019; Durie et al., 2011; Galbán et al., 2008).

The anti-angiogenic function of VASH1 may appear inconsistent with its ability to activate the IRES of an angiogenic factor. However, our data also suggest that VASH1 might be an activator or an inhibitor of (lymph)angiogenic factor mRNA translation. Such a double role may explain the unique dual ability of VASH1 to inhibit angiogenesis and to promote endothelial cell survival (Sato, 2015; Miyashita et al., 2012). This could result from the existence of different VASH1 isoforms of 44 kDa, 42 kDa, 36 kDa, 32 kDa and 27 kDa, resulting from alternative splicing and/protein processing (Kishlyansky et al., 2010; Kern et al., 2008; Sato, 2013; Sonoda et al., 2006). Interestingly, p42 and p27 are the main isoforms expressed in heart, where the p44 is undetectable (Kishlyansky et al., 2010; Sonoda et al., 2006). One can expect that the ITAF function is carried by p42, which contains the two predicted RNA-binding domains (Figure 5C, Figure 5—figure supplement 1). VASH1 has been observed in both the nucleus and the cytoplasm, and no striking nucleocytoplasmic relocalization is visible in response to hypoxia, whereas other ITAFs (such as hnRNPA1 or nucleolin) shuttle to the cytoplasm upon stress (Godet et al., 2019; Morfoisse et al., 2016; Lewis et al., 2007; Cammas et al., 2007; Dobbyn et al., 2008). Interestingly, VASH1, appears as foci whose size increases in hypoxia, suggesting that it could be partly translocated to stress granules. This translocation has been reported for other ITAFs, such as hnRNPA1 and polypyrimidine-tract-binding protein (PTB) (Godet et al., 2019; Borghese and Michiels, 2011; Guil et al., 2006).

VASH1 impact on translational control can be positive or negative

Among the IRESs analyzed in the present study, the FGF1 IRES is the only one strongly regulated by VASH1 in hypoxia. However, VASH1 was also bound to the EMCV IRES in the BIA-MS experiment, and calculation of affinity constants does not reveal significant differences in affinity for FGF1, VEGFA or EMCV IRES. This apparent inconsistency finds an explanation if one considers the effect of VASH1 in normoxia: IRESs tend to be activated upon VASH1 depletion, significantly so for VEGFD and EMCV IRESs. Such data suggest that VASH1 binding is probably not specific to a given IRES, but instead that different VASH1 partners are recruited in the IRESome and result in positive or negative effects of this ITAF. The hypothesis of a dual role for VASH1 in translational control is confirmed by the effect of VASH1 depletion on the translatome: recruitment into polysomes is affected negatively or positively for 60–70% of mRNAs, both in normoxia and in hypoxia. Although the RNA-binding ability of VASH1 has been clearly shown in the present study, we cannot affirm that VASH1 impact is direct for all of these mRNAs. Nevertheless, a dual role of activator and inhibitor has been reported for more than ten other ITAFs. Our hypothesis thus remains that the key to the regulation of IRES activity by ITAFs is not RNA-binding specificity but rather IRESome multi-partner composition (Godet et al., 2019).

Materials and methods

Key resources table
Reagent type
(species) or resource		DesignationSource or referenceIdentifiersAdditional
information
Gene (firefly, Photinus)Luc+PromegaModified firefly luciferase
Gene (Renilla reniformis)LucRPromegaRenilla luciferase
Strain, strain background (Escherichia coli)TOP 10Thermofisher ScientificTransformation-competent cells
(genotype : F– mcr A Δ (mrr –hsd RMS–mcr BC) φ 80lac ZΔ M15 ΔlacX 74 rec A1 ara D139 Δ (araleu) 7697 gal U gal K rps L
(StrR) end A1 nup G)
Cell line (Homo sapiens)293FTInvitrogenR700-07High-transfection performance for lentivector production
Cell line (Homo sapiens)HT1080ATCCCCL-121
Cell line (Mus musculus)HL-1William C. Claycomb (Claycomb et al., 1998)Cardiomyocyte cell line with beating phenotype
Recombinant DNA reagentpTRIP-CRHL+Dryad,
Supplementary file 8A (Morfoisse et al., 2014)		Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by an intergenic palindromic sequence
Recombinant DNA reagentpTRIP-CRF1AL+Sequence available on Dryad, (Martineau et al., 2004; Ainaoui et al., 2015)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human FGF1 IRES A
Recombinant
DNA reagent		pTRIP-CRFL+Sequence available on Dryad, (Créancier et al., 2000)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human FGF2 IRES
Recombinant DNA reagentpTRIP-CRVAL+Sequence available on Dryad, (Huez et al., 1998)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human VEGFA IRESa
Recombinant
DNA reagent		pTRIP-CRVBL+Sequence available on Dryad, (Huez et al., 1998)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human VEGFA IRESb
Recombinant DNA reagentpTRIP-CRhVCL+Sequence available on Dryad, (Morfoisse et al., 2014)Bicistronic SIN lentivector
construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human VEGFC IRES
Recombinant DNA reagentpTRIP-CRhVDL+Sequence available on Dryad, (Morfoisse et al., 2016)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human VEGFD IRES
Recombinant DNA reagentpTRIP-CRMP2L+Sequence available on Dryad, (Nanbru et al., 1997)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the human c-myc IRES
Recombinant
DNA reagent		pTRIP-CREL+Sequence available on Dryad, (Créancier et al., 2000)Bicistronic SIN lentivector construct with the CMV promoter controlling expression of LucR and Luc+ separated by the EMCV IRES
Recombinant DNA reagentpCMV-dR8.91AddgeneTrans-complementing plasmid containing the lentiviral protein genes (gag, pol, rev…)
Recombinant DNA reagentpCMV-VSV-GAddgene#8454Trans-complementing plasmid containing VSV G envelope protein gene
Transfected construct (mouse)Acell SMARTpool targeting VASH1 (siVASH1)DharmaconsiRNA targeting VASH1 (Supplementary file 8B)
Transfected construct (mouse)Control SMARTpoolDharmaconScramble siRNA (Supplementary file 8B)
Sequence-based reagentDeltagene assayFluidigm Corporation, oligonucleotide sequences available in Supplementary file 6PCR array with 96 oligonucleotides primer couples
AntibodyMouse monoclonal anti-VASH-1AbcamEPR17420Jess Western 1:100
AntibodyMouse monoclonal anti-VASH-1Abcamab176114IC 1:50
AntibodyMouse monoclonal anti-FGF1-1AbcamEPR19989Jess Western 1:25
AntibodyMouse monoclonal anti-p21Santa Cruzsc-6546 (F5)Jess Western 1:10
AntibodyRabbit polyclonal anti eIF2αCell Signaling Technology9721Jess Western 1:50
AntibodyMouse monoclonal anti-phospho-eIF2αCell Signaling Technology2103Jess Western 1:50
AntibodyRabbit polyclonal anti-4EBP-1Cell Signaling Technology9452Jess Western 1:50
AntibodyRabbit polyclonal anti-phospho-4EBP-1Cell Signaling Technology9451Jess Western 1:50
AntibodySecondary-HRP (ready to use rabbit ‘detection module’)Protein SimpleDM-001Jess Western
AntibodyDonkey anti-rabbit alexa488Jackson Immunoresearch711-545-152Secondary
antibody for IC
Software, algorithmPRISMGraphpadStatistical analysis

Lentivector construction and production

Request a detailed protocol

Bicistronic lentivectors coding for the renilla luciferase (LucR) and the stabilized firefly luciferase Luc+ (called LucF in the text) were constructed from the dual luciferase lentivectors described previously, which contained Luc2CP (Morfoisse et al., 2014; Morfoisse et al., 2016). The LucR gene used here is a modified version of LucR in which all the predicted splice donor sites have been mutated. The cDNA sequences of the human FGF1, -2, VEGFA, -C, -D, c-myc and EMCV IRESs were introduced between the first (LucR) and the second cistron (LucF) (Vagner et al., 1995; Nanbru et al., 1997; Prats et al., 2013). IRES sequence sizes are: 430 nt (FGF1), 480 nt (FGF2), 302 nt (VEGFAa), 485 nt (VEGFAb), 419 nt (VEGFC), 507 nt (VEGFD), 363 nt (c-myc), and 640 nt (EMCV) (Morfoisse et al., 2014; Morfoisse et al., 2016; Huez et al., 1998; Martineau et al., 2004; Vagner et al., 1995; Nanbru et al., 1997). The two IRESs of the VEGFA have been used and are called VEGFAa and VEGFAb, respectively (Huez et al., 1998). The hairpin negative control contains a 63 nt long palindromic sequence cloned between LucR and LucF genes (Supplementary file 8A). This control has been successfully validated in previous studies (Morfoisse et al., 2014; Créancier et al., 2000). The expression cassettes were inserted into the SIN lentivector pTRIP-DU3-CMV-MCS vector described previously (Prats et al., 2013). All cassettes are under the control of the cytomegalovirus (CMV) promoter. All vector sequences are available on Dryad (Ape format). Plasmid construction and amplification was performed in the bacteria strain TOP10 (Thermofisher Scientific, Illkirch Graffenstaden, France).

Lentivector particles were produced using the CaCl2 method by tri-transfection with the plasmids pCMV-dR8.91 and pCMV-VSVG, CaCl2 and Hepes-buffered saline (Sigma-Aldrich, Saint-Quentin-Fallavier, France) into HEK-293FT cells. Viral supernatants were harvested 48 hr after transfection, passed through 0.45 μm PVDF filters (Dominique Dutscher SAS, Brumath, France) and stored in aliquots at −80°C until use. Viral production titers were assessed on HT1080 cells with serial dilutions of a lentivector expressing green fluorescent protein (GFP) and scored for GFP expression by flow cytometry analysis on a BD FACSVerse (BD Biosciences, Le Pont de Claix, France).

Cell culture, transfection and transduction

Request a detailed protocol

293FT (Invitrogen R700-07) and HT1080 (ATCC CCL-121) cells were provided by Invitrogen (Villebon sur Yvette, France) and ATCC (Manassas, VA, USA), respectively. The 293FT cell line is derived from human embryonic kidney cells transformed by the simian virus 40 (SV40) large T antigen. This cell line is ideal for the production of high titers of lentivectors. HT1080 is a human transformed line expressing activated N-ras oncogene. It was used only for lentivector titration.

The two cell lines were cultured in DMEM-GlutaMAX + Pyruvate (Life Technologies SAS, Saint-Aubin, France), supplemented with 10% fetal bovine serum (FBS), and MEM essential and non-essential amino acids (Sigma-Aldrich).

Mouse atrial HL-1 cardiomyocytes were a kind gift from Dr. William C. Claycomb (Department of Biochemistry and Molecular Biology, School of Medicine, New Orleans) (Claycomb et al., 1998). HL-1 cells are derived from a tumor of a transgenic mouse in which expression of the SV40 large T antigen was targeted to atrial cardiomyocytes. These highly differentiated cardiomyocyte HL-1 cells can be cultured and maintain their cardiac (beating) phenotype. The authentication method is observation of the beating phenotype. As soon as the phenotype is lost, it is necessary to start again with cells from an earlier passage.

All the cell lines were tested negative for mycoplasma contamination every three months. None of them is on the list of the commonly misidentified cell lines maintained by the International Cell Line Authentication Committee.

HL-1 cells were cultured in Claycomb medium containing 10% FBS, penicillin/streptomycin (100 µg/mL-100µg/mL), 0.1 mM norepinephrine, and 2 mM L-glutamine. Cell culture flasks were pre-coated with a solution of 0.5% fibronectin and 0.02% gelatin 1 hr at 37°C (Sigma-Aldrich). To keep the HL-1 phenotype, cell culture was maintained as previously described (Claycomb et al., 1998). For hypoxia, cells were incubated at 37°C at 1% O2. HL-1 cardiomyocytes were transfected by siRNAs as follows: one day after being plated, cells were transfected with 10 nM of small interference RNAs from Dharmacon Acell SMARTpool, targeting VASH1 (siVASH1) or as a non-targeting siRNA control (siControl), using Lipofectamine RNAiMax (Invitrogen) according to the manufacturer’s recommendations, in a media without penicillin-streptomycin and norepinephrine. Cells were incubated for 72 hr at 37°C with siRNA (siRNA sequences are provided in Supplementary file 8B).

For lentivector transduction, 6.104 HL-1 cells were plated into each well of a six-well plate and transduced overnight in 1 mL of transduction medium (OptiMEM-GlutaMAX, Life Technologies SAS) containing 5 µg/mL protamine sulfate in the presence of lentivectors (MOI 2). A lentivector expressing GFP was used as a transduction control. GFP-positive cells were quantified 48 hr later by flow cytometry analysis on a BD FACSVerse (BD Biosciences). HL-1 cells were transduced with 80% efficiency. siRNA treatment of transduced cells was performed 72 hr after transduction (and after one cell passage).

To measure protein half-life, HL-1 cardiomyocytes were treated with cycloheximide (InSolution CalBioChem) diluted in PBS to a final concentration of 10 µg/mL in well plates. Time-course points were taken by stopping cell cultures after 0 hr, 4 hr, 6 hr, 8 hr, 16 hr or 24 hr of incubation and subsequent capillary Western analysis of cell extracts.

Reporter activity assay

Request a detailed protocol

For reporter lentivectors, luciferase activities in vitro and in vivo were measured using a Dual-Luciferase Reporter Assay (Promega, Charbonnières-les-Bains, France). Briefly, proteins from HL-1 cells were extracted with Passive Lysis Buffer (Promega France). Quantification of bioluminescence was performed with a luminometer (Centro LB960, Berthold, Thoiry, France).

Capillary electrophoresis and Jess simple western

Request a detailed protocol

Diluted protein lysate was mixed with fluorescent master mix and heated at 95°C for 5 min. 3 µL of protein mix containing protein normalization reagent, blocking reagent, wash buffer, target primary antibody (mouse anti-VASH-1 [Abcam EPR17420] diluted 1:100; mouse anti-FGF1 [Abcam EPR19989] diluted 1:25; mouse anti-P21 [Santa Cruz sc-6546 (F5)] diluted 1:10; rabbit anti eIF2α [Cell Signaling Technology 9721] diluted 1:50; mouse anti-phospho-eIF2α [Cell Signaling Technology 2103] diluted 1:50; rabbit anti-4EBP-1 [Cell Signaling Technology 9452] diluted 1:50; rabbit anti-phospho-4EBP-1 [Cell Signaling Technology 9451] diluted 1:50), secondary-HRP (ready-to-use rabbit ‘detection module’ [Protein Simple DM-001]), and chemiluminescent substrate were dispensed into designated wells in a manufacturer-provided microplate. The plate was loaded into the instrument (Jess, Protein Simple) and proteins were drawn into individual capillaries on a 25 capillary cassette (12–230 kDa) (SM-SW001). Data were analyzed using the compass software provided by the manufacturer. Normalization reagent allowed the detection of total proteins in the capillary through the binding of amine group by a biomolecule, and removed housekeeping proteins that can cause inconsistent and unreliable expression. No loading control is required with the Jess technology. The graphs in the figures show chemiluminescence values before normalization.

RNA purification and cDNA synthesis

Request a detailed protocol

Total RNA extraction from HL-1 cells was performed using TRIzol reagent according to the manufacturer’s instructions (Gibco BRL, Life Technologies, NY, USA). RNA quality and quantification were assessed using an Xpose spectrophotometer (Trinean, Gentbrugge, Belgium). RNA integrity was verified with an automated electrophoresis system (Fragment Analyzer, Advanced Analytical Technologies, Paris, France).

500 ng RNA was used to synthesize cDNA using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Villebon-sur-Yvette, France). Appropriate no-reverse transcription and no-template controls were included in the PCR array plate to monitor potential reagent or genomic DNA contaminations, respectively. The resulting cDNA was diluted 10 times in nuclease-free water. All reactions for the PCR array were run in biological triplicates.

qPCR array

Request a detailed protocol

The DELTAgene Assay was designed by the Fluidigm Corporation (San Francisco, USA). The qPCR-array was performed on BioMark with the Fluidigm 96.96 Dynamic Array, following the manufacturer’s protocol (Real-Time PCR Analysis User Guide PN 68000088). The list of primers is provided in Supplementary file 6. A total of 1.25 ng of cDNA was preamplified using PreAmp Master Mix (Fluidigm, PN 100–5580, 100–5581; San Francisco, USA) in the plate thermal cycler at 95°C for 2 min, 10 cycles at 95°C for 15 s and 60°C for 4 min. The preamplified cDNA was treated by endonuclease I (New England BioLabs, PN M0293L; Massachusetts, USA) to remove unincorporated primers.

The preamplified cDNA was mixed with 2x SsoFast EvaGreen Supermix (BioRad, PN 172–5211; California, USA), 50 μM of mixed forward and reverse primers and sample Loading Reagent (Fluidigm, San Francisco, USA). The sample was loaded into the Dynamic Array 96.96 chip (Fluidigm San Francisco, USA). The qPCR reactions were performed in the BioMark RT-qPCR system. Data were analyzed using the BioMark RT-qPCR Analysis Software Version 2.0.

18S rRNA was used as a reference gene and all data were normalized on the basis of 18S rRNA level. Hprt was also assessed as a second reference gene but was not selected as its level was not stable during hypoxia. Relative quantification (RQ) of gene expression was calculated using the 2–ΔΔCT method. When the RQ value was inferior to 1, the fold change was expressed as −1/RQ. The oligonucleotide primers used are detailed in Supplementary file 6.

Polysomal RNA preparation

Request a detailed protocol

HL-1 cells were cultured in 150 mm dishes. 15 min prior to harvesting, cells were treated with cycloheximide at 100 μg/ml. Cells were washed three times in PBS cold containing 100 μg/mL cycloheximide and scraped in the PBS/cycloheximide. After centrifugation at 3000 rpm for 2 min at 4°C, cells were lysed by 450 μl hypotonic lysis buffer (5 mM Tris-HCL [pH7.5]; 2.5 mM MgCl2; 1.5 mM KCl). Cells were centrifuged at 13,000 rpm for 5 min at 4°C, before the supernatants were collected and loaded onto a 10–50% sucrose gradient. The gradients were centrifuged in a Beckman SW40Ti rotor at 39,000 rpm for 2.5 hr at 4°C without brake. Fractions were collected using a Foxy JR ISCO collector and UV optical unit type 11. RNA was purified from pooled heavy fractions containing polysomes (fractions 19–27), as well as from cell lysate, before gradient loading.

Preparation of biotinylated RNA

Request a detailed protocol

The FGF1, VEGFA or EMCV IRESs was cloned in pSCB-A-amp/kan plasmid (Agilent) downstream from the T7 sequence. The plasmids were linearized and in vitro transcription was performed with a MEGAscript T7 kit (Ambion), according to the manufacturer’s protocol, in the presence of Biotin-16-UTP at 1 mM (Roche), as previously described (Ainaoui et al., 2015). The synthesized RNA was purified using an RNeasy kit (Qiagen).

BIA-MS experiments

Request a detailed protocol

BIA-MS studies based on surface plasmonic resonance (SPR) technology were performed on a BIAcore T200 optical biosensor instrument (GE Healthcare), as described previously (Morfoisse et al., 2016; Ainaoui et al., 2015). Immobilization of biotinylated IRES RNAs was performed on a streptavidin-coated (SA) sensorchip in HBS-EP buffer (10 mM Hepes [pH 7.4], 150 mM NaCl, 3 mM EDTA, 0.005% surfactant P20) (GE Healthcare). All immobilization steps were performed at a flow rate of 2 µl/min with a final concentration of 100 µg/ml.

Binding analyses were performed with normoxic or hypoxic cell protein extracts at 100 µg/ml over the immobilized IRES RNA surface for 120 s at a flow rate of 30 µl/min. The channel (Fc1) was used as a reference surface for non-specific binding measurements. The recovery wizard was used to recover selected proteins from cell protein extracts. This step was carried out with 0.1% SDS. Five recovery procedures were performed to get amounts of proteins sufficient for MS identification.

Eluted protein samples from BIA experiment were digested in gel with 1 µg of trypsin (sequence grade, Promega) at 37°C. Peptides were then subjected to LC-MS/MS analysis. The peptide mixtures were loaded on a YMC-Triart C18 150 × 300 µm capillary column (particle diameter 3 µm) connected to a RS3000 Dionex HPLC system. The run length gradient (acetonitrile and water) was 30 min. Then, on the AB Sciex 5600+ mass spectrometer, data were acquired with a data-dependent analysis. Data were then loaded on Mascot software (Matrix Science) that attributes peptide interpretations to MS/MS recorded scans. The higher the score, the lower the probability of a false positive (a score of 20 corresponds to a 5% probability of a false positive).

Surface plasmon resonance assays

Request a detailed protocol

For kinetic analysis, immobilization of biotinylated FGF1 IRES RNA was performed on a streptavidin-coated (SA) sensorchip in HBS-EP buffer (10 mM Hepes [pH 7.4], 150 mM NaCl, 3 mM EDTA, 0.005% surfactant P20) (GE Healthcare). The immobilization step was performed at a flow rate of 2 µl/min with a final concentration of 100 µg/ml. The total amount of immobilized FGF1 IRES RNA was 1500 RU.

Binding analyses were performed with recombinant protein VASH1 (Abnova H00022846-P01) at 100 µg/ml over the immobilized FGF1. This recombinant VASH1 contains the 27-kDa N-terminal part of the protein coupled to glutathione S-transferase. The channel (Fc1) was used as a reference surface for non-specific binding measurements.

A Single-Cycle Kinetics (SCK) analysis to determine association, dissociation and affinity constants (ka, kd, and KD respectively) was carried out by injecting different protein concentrations (16.25–300 nM). Binding parameters were obtained by fitting the overlaid sensorgrams with the 1:1. Langmuir binding model of the BIA evaluation software version 3.0.

Immunocytology

Request a detailed protocol

Cells were plated on glass coverslips and incubated for 4 hr of normoxia or hypoxia. They were fixed with cold methanol at −20°C for 5 min, washed three times with PBS, and permeabilized for 1 min with 0.1% Triton. Then, cells were incubated for 5 min with blocking solution (1% FBS, 0.5% BSA) and 30 min with anti-VASH1 antibody (1/50; abcam ab176114) and Alexa 488 conjugated anti-mouse secondary antibody. Images were acquired with a LSM780 Zeiss confocal microscope, camera lens x60 with Z acquisition of 0.36 μM. A single plan is shown Figure 6C.

Imaris software was used to represent vasohibin staining in Figure 6C. To differentiate vasohibin in the nucleus and cytoplasm, the nucleus was delimitated with DAPI staining and all vasohibin foci in the nucleus are shown in purple, whereas those in the cytoplasm are shown in green.

Using Imaris software, the mean number of vasohibin foci was counted and the volume of vasohibin foci was quantified, a threshold was applied and all particles above 0.5 μm3 were selected and quantified.

Statistical analysis

Request a detailed protocol

All statistical analyses were performed using two-tailed Student's t-tests (Figure 3), Mann-Whitney tests (Figure 4 and Figure 7) or one-way Anova with Tukey’s comparisons test (Figure 6), *p<0.05, **p<0.01, ***<0.001, ****<0.0001. Data are expressed as mean ± standard deviation.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. Lentivector plasmid complete maps and sequences are available on Dryad.

The following data sets were generated
    1. Hantelys F
    2. Godet A
    3. David F
    4. Tatin F
    5. Renaud-Gabardos E
    6. Pujol F
    7. Diallo L
    8. Ligat L
    9. Henras A
    10. Sato Y
    11. Parini A
    12. Lacazette E
    13. Garmy-Susini B
    14. Prats A
    15. Ader I
    (2019) Dryad Digital Repository
    Data from: Vasohibin1, a new IRES trans-acting factor for induction of (lymph)angiogenic factors in early hypoxia.
    https://doi.org/10.5061/dryad.2330r1b

References

Decision letter

  1. Nahum Sonenberg
    Reviewing Editor; McGill University, Canada
  2. James L Manley
    Senior Editor; Columbia University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The authors document the function of internal ribosome entry sites (IRESs) in mRNA translational control in cardiomyocytes under hypoxic conditions. They show that IRESes in several mRNAs encoding FGF and VEGF family members function in hypoxic cells. They also discovered that a surprisingly new IRES-trans-acting factor (ITAF) vasohibin (VASH1) is required for IRES-dependent translation, as VASH1 was reported earlier to exhibit anti-angiogenic activity in vivo in endothelial cells. Their paper suggests that the IRES-ITAF composition varies at different stages of hypoxia, engendering the sequential production of angiogenic factors required to form new functional vessels in the ischemic heart. The paper is highly relevant to the understanding of vascular biology, and pathologies in which hypoxia plays a major role, such as cardiovascular diseases and cancer.

Decision letter after peer review:

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Vasohibin1, a new IRES trans-acting factor for sequential induction of angiogenic factors in hypoxia" for consideration by eLife. Your article has been reviewed by a Senior Editor, a Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work cannot be considered further for publication in eLife.

The three reviewers agree that the major problem is the lack of sufficient detail and rigor in this work and manuscript. The polysomes need to be quantified (P/M ratio) or repeated since the amount of lysate loaded is not identical. You should report raw numbers for normalized values, including statistical analysis and showing all data when possible rather than representative data. The manuscript is overly speculative given the data presented. The data do not provide sufficient support to the conclusion of waves of IRES activation or that Vash1 binding and relocalization are regulating IRES activity. These conclusions either have to be supported by more data or dramatically re-written. Thus, there is a significant amount of work that is required to render this paper accepted for publication in eLife. Because it is not clear whether the experiments can be performed in 2 months, we cannot accept the paper but would be willing to consider a new submission that successfully addresses all the issues raised.

Reviewer #1:

The manuscript by Prats et al., aims to better understand the role of IRESs in cardiomyocytes under hypoxic conditions. They show transcriptomic and polysome analysis that a lot of mRNAs that encode for proteins important for angiogenesis and lymph-angiogenesis are translationally regulated. They go on to show that many of these mRNAs have IRES activity, which is temporally regulated during hypoxia. Then they identify one ITAF, VASH1, that is required for FGF1 IRES activity. Overall, this is a very nice study that makes a significant contribution to our understanding of hypoxia in immortalized muscle cells. However, the biggest issue with this manuscript is the lack of details. Many of the experiments are not explained well enough in the methods or the figure legend for the reader to be confident that they are interpreting the results correctly.

Essential revisions:

It would be useful for the authors to justify why they chose 4, 8 and 24 hours for their time course. It would be nice to at least see how the IRES activity changed over a more detailed time course from 30 minutes to 24 hours having 10-15 time points. This would provide a better picture of the timing of the IRES regulation. For example, if hypoxia is translationally regulated why wouldn't we expect differences in 30 minutes or less? It would also be more compelling that there were differences in the timing of translation of these IRES containing mRNAs if there was more of a trend especially since the differences are so small (which doesn't mean that they aren't real). Also, given that the discussion focuses on how in many tumor cell lines that these IRESs are only activated after 24 hours of hypoxia, it would be informative to see if time points later than 24 hours resulted in a more robust activation of IRES activity. This would be necessary to support one of their major conclusions that the timing of IRES-mediated translation is regulated during hypoxia.

Figure 2A the authors claim that "The polysome profile displayed, as expected, a strong decrease of global translation (Figure 2A)." However, if there was a decrease in polysomes then there should be a corresponding increase in free subunits (40S and 60S), which is not seen. Rather their data show poor translational efficiency during normoxia and the reduced polysome peaks appear to be due to loading of less lysate. With different levels of lysate on the gradient the only definitive way to show that there is a significant reduction is to calculate the polysome (2-mer and up) to monosome (80S peak) ratio. The P/M should be shown for both normoxia and hypoxia polysomes.

In Figure 2 it is not clear how they calculated the fold change of the polysome associated mRNA under hypoxia. This is not adequately described in the figure legend and is not in the methods at all under Polysomal RNA preparation. Which fractions were collected for polysomes? Which fractions were free RNA? Whether all the fractions in the gradient were measured, which is important since equivalent levels of lysate was not used. How the statistics were calculated. The specific statistical test used should be in the figure legend rather than listing of all tests used in the manuscript in the methods. What is meant by "normalized to normoxia as in Figure 1" (which isn't explained in Figure 1 either).

In Figure 3 it is not clear again how the Fold-change in polysomal mRNA was calculated. Comparing it to polysome associated in normoxia? Or looking at the shift from monosomes to polysomes under hypoxic conditions. The first has issues with sample differences and these would need to be adequately addressed. The latter approach must be clearly detailed. Alternatively, if only the polysomes were examined then this would have issues with sample variation since generally polysomes are internally controlled and if only a fraction of the polysomes are assayed then it becomes impossible to know what fraction of the mRNA was polysome associated.

In Figure 4 they report that representative assays are shown. First the raw data needs to be reported in the supplementary material for all the replicates. This would include an empty vector with no IRES to show the background levels. Second, it is not clear why they are reporting representative data since it is a ratio of Fluc/Rluc which in theory should correct for differences in transduction efficiency. There is a significant concern that the differences observed are due to changes in Rluc values going down during hypoxia rather than a real increase in Fluc values. This should be cleared up by showing the raw data.

Table 2 the legend says "Relative quantification (RQ) of gene expression in hypoxia was calculated using the 2-ΔΔCT method (polysomal RNA/free RNA normalized to normoxia)." Does this mean that the data was normalized to normoxia either with or without siRNA KD of VASH1? Or was the KD data normalized to normoxia with VASH1 present in the cells? This should be clarified

The argument that the authors make for mRNAs that are translationally up-regulated by the VASH1 ITAF need VASH1 to re-localize to the nucleus to form the "IRESome". However, there is no data to show that preventing re-localization of VASH1 to the nucleus affects FGF1 IRES activity. It seems just as likely that the nuclear localization of the VASH1 may be for another function. Also, given the increase in VASH1 protein under hypoxia and its translational up-regulation, its overall nuclear to cytoplasmic distribution may remain unchanged but is more apparent in the nucleus at the higher expression levels. There is clearly VASH1 in the nucleus under normoxic conditions.

The authors should address why their reporter data does not match with their polysome data in that the reporter data shows the VEGFAa and VEGFAb IRESs were not increased above normoxic levels at 4 hours hypoxia (Figure 4C) whereas the polysome data shows they clearly are translationally upregulated (Figure 2 and Figure 3B). It is possible that they would see more dramatic changes in IRES activity of the reporters if they used a promoter such as the SV40 promoter, which is expressed at more physiological levels. There is some concern that using the higher expressing CMV promoter for these cellular IRESs reporters could saturate ITAFs therefore muting differences in IRES translation under hypoxic conditions for their reporters.

The methods for the real-time qPCR experiments is not clear. See the MIQE publication for best practices. We don't know how many replicates were measured, whether there were multiple dilutions of the sample measured, what the efficiency for the reactions were, which becomes critical given the small differences that were reported between samples. See "The MIQE Guidelines: Minimum Information for Publication of Quantitative Real-Time PCR Experiments".

Reviewer #2:

Hantelys and colleagues analyse changes in transcriptome and polysome association of mRNAs encoding lymphangiogenic factors during hypoxia. They observe differential activation of IRES-dependent translation along the hypoxic response and propose VASH1 as a protein binding and regulating the activity of certain IRESes. However, the manuscript at its present point is not convincing. There are several important controls missing in this study which are needed to support the conclusions:

1) Figure 2A: A decrease in polysomes should be accompanied by an increase in sub-polysomes or monosomal fractions. This does not seem to be the case of the profile shown here, suggesting that less material has been loaded in the gradient.

2) Table 3:

- What does the score mean and what would be a "good score"?

- Most proteins in this Table are detected with just one peptide. Proteins that are usual contaminants in affinity chromatography experiments (e.g. serum albumin) are also detected with one peptide here. How are true interactions distinguished from background?

- From the data in Table 3, the choice of VASH1 as a potential ITAF seems aleatory because of the low level of detection and the fact that there is no difference between 4 hours (where the FGF1 IRES is active) and 8 hours (where the IRES is inactive). This protein is detected with only one peptide in all instances, so the "selective binding depending on the IRES and stress condition" alluded to by the authors in the main text is difficult to believe.

3) Figure 5: The RNA binding constant of VASH1 for FGF1 IRES is somewhat high and the RBDs are only predicted. Both RNA binding and the assignment of VASH RBDs should be backed-up by controls. For example, how does binding to VEGF IRES looks like? Is binding to FGF1 IRES competed by FGF1 and EMCV IRESes but not by VEGF IRES? Does mutation of the predicted RBDs in VASH1 eliminate binding? Importantly, does VASH1 bind to endogenous FGF1 mRNA?

4) Figure 6: Please, show a Western blot with and without VASH1 depletion to ensure that VASH1 antibody only detects the expected protein(s). I do not see a perinuclear localization pattern, but rather a cytoplasmic pattern.

5) Figure 7C: Given that the activity of the various IRESes in the bicistronic assay is weak (2-fold maximum), it is important to ensure that the reduction after VASH1 depletion is significant and is not affected by the variability inherent to the assays. For this reason, rather than showing one representative triplicate experiment and use standard error of the mean to measure significance, it is important to use the full set of biological replicates and show standard deviation. The same could be applied to Figure 4B-D.

Also regarding Figure 7C, how do the authors explain the increase in VEGFA IRES activity upon VASH1 depletion if VASH1 does not bind to VEGFA IRES? The suggested inhibitory effect of VASH1 is more than discussible.

6) How does depletion of VASH1 affect the polysomal distribution of FGF1 and VEFGA mRNAs in hypoxia? Table 2 shows a comparison of polysomes in hypoxia (4 hours) with siVASH1 polysomes in normoxia, while the correct comparison would be siControl polysomes in hypoxia (4 hours) with siVASH1 polysomes in hypoxia (4 hours).

7) The Discussion section is highly speculative. The existence of "waves" of IRES-dependent regulation cannot be supported by the analysis of just 2-4 IRESes representing about 5% of the full set of translationally activated mRNAs under hypoxia. Furthermore, the physiological relevance of the potential regulation of IRES-dependent translation by VASH1 cannot be inferred by the established roles of VASH1, because each factor regulates multiple targets by different mechanisms. In addition, whether nuclear localization of VASH1 has any relevance for the function described here is an open question. The inference on regulons is too far-fetched given that there is only one member of each proposed regulon.

Reviewer #3:

In this manuscript the authors show that the FGF1 IRES functions in hypoxic cells, using a pathophysiological relevant system, and that vasohinbin1 is required for IRES-dependent translation. It is of interest that the authors have studied IRESs in models of hypoxia other than in tumour cells and the data provide new insights into how these regulatory RNA elements function. In general, the experiments have been well performed, but a couple of changes to the way in which the data are presented would help with the clarity of the MS, in addition to a small amount of additional experimentation.

Figure 2:

While the data show that there are certainly less "heavy" polysomes in the hypoxic cells it would be better if the areas under the curves were calculated so that these data can be presented additionally as a polysome:subpolysome ratio to provide a stronger indication of the degree of initiation changes. This would also allow the extent of variation between experimentation to be observed. In addition, the authors need to carry out Met labelling or use some other measure to show that global protein synthesis rates are decreasing under these conditions. This should be accompanied by western analysis to show changes in for example, 4EBPs, eIF2α.

Figure 3

The data show fold-changes in polysome mRNA localisation and in addition changes in transcription. Since the transcription and translation data have been obtained using arrays and, it is important to show how these transcripts vary across the gradients in hypoxic cells to confirm these findings, for a couple of mRNAs. Is it possible to calculate changes in translational efficiency from these data taking into account the transcriptional variation? This information would be very useful.

Figure 4

The data show how the IRES function changes, but it is important to show how these data correlate with changes in expression of the corresponding proteins. For example, is there an increase in VEGF of 1.9 fold after 8 hours of hypoxia? If the half-lives of these proteins are too long, pulse IP may need to be used.

Figure 6.

It is essential to measure the half-life of VASH1. Another explanation for the protein remaining expressed, even though the mRNA is decreasing, is that it is very stable with a long half-life. Therefore, it would not be turned over sufficiently during the course of the experiment to see a real difference.

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

Thank you for submitting your article "Vasohibin1, a new IRES trans-acting factor for induction of (lymph)angiogenic factors in early hypoxia" for consideration by eLife. Your article has been reviewed by James Manley as the Senior Editor, a Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The authors investigated the role of IRESs in cardiomyocytes under hypoxic conditions. They show that the FGF1 IRES functions in hypoxic cells and that vasohinbin1 is required for IRES-dependent translation. They also show by polysome analysis that many mRNAs that encode for proteins important for angiogenesis and lymphangiogenesis are translationally regulated, many contain an IRES activity, which is temporally regulated during hypoxia.

Essential revisions:

The three original reviewers of your paper agree that you performed many additional important experiments and that the paper has improved considerably. However, as you can see two reviewers argue that some key issues remain to be addressed before the paper can be recommended for publication in eLife. These concerns were discussed and agreed on by all reviewers. In particular, it is critical to show that protein expression of endogenous VASH1 changes as the paper claims. You need to carry out pulse labeling IP of VASH1 to convincingly show changes in synthesis. You also ought to provide evidence that VASH1 specifically binds to the endogenous FGF-1 IRES. Another important issue that was raised by reviewer #2 concerns the statistical analysis of the data. The reviews of the three referees are attached so that you can read the detailed comments.

Reviewer #1:

This is an interesting study where it is shown that in hypoxic cells post-transcriptional control plays a major role in gene expression regulation. In particular the translation of IRES-containing mRNAs is upregulated and this is in part mediated by VASH1. The authors have carried out all additional experiments requested, and this has strengthened the manuscript. In my opinion this manuscript is now suitable for publication.

Reviewer #2:

This is a revised manuscript from Hantleys et al., entitled "Vasohibin1, a new IRES trans-acting factor for induction of (lymph)angiogenic factors in early hypoxia". The authors have been very responsive to the reviewer comments and the revised manuscript is significantly improved. Importantly, they showed the raw IRES luciferase values and included a control that would eliminate readthrough from the first to the second cistron (hairpin). Also, this negative control had 3 logs lower RLU values, which shows that they are measuring RLU activity well above background. The findings that are significant and novel include the study of IRESs in immortalized cells rather than transformed cells. Their work used immortalized cardiomyocytes to show that IRESs are induced during hypoxia. Interestingly, they found upregulation of hypoxic IRESs much earlier in these cell lines compared to what others have shown in transformed cells suggesting that cancer cells may be more resistant to hypoxia than normal cells. Yet, given that the cells they used are still immortalized and not primary cells, we still do not know how primary cells use IRESs during hypoxia. They show that several mRNAs become polysome associated following hypoxia and in particular they show that FGF1 IRES activity increases following induction of hypoxia. They also identify an ITAF, VASH1, that binds to IRESs in the nanomolar range and that FGF1 IRES activity depends on this ITAF during hypoxia.

However, there are a few things that reduce enthusiasm such as the Fgf1 western which doesn't look like there is an increase in protein levels under hypoxic conditions. Also, the IRES activities increase during hypoxia, but only by 1.2 to 1.7-fold. Is this sufficient enough for a biological response? Perhaps this is why the western didn't show a significant change in FGF1 protein levels or perhaps the increase in IRES activity is muted simply due to using immortalized cells or a promoter that overexpresses the dicistronic RNA. It is difficult to say. Even the ITAF, VASH1, which they reported bound to EMCV and FGF1 IRESs but not VEGFA IRES still had similar binding Kd for all of the IRESs. Overall, many of the conclusions are based on small effects or minor discrepancies that weaken the main conclusions, however, it is possible that these small changes in IRES activity are sufficient to alter cellular programs.

The authors reference a Figure 3C when there isn't a Figure 3C.

One of the key figures in the paper that supports a major conclusion that IRES activity leads to increased Fgf1 protein levels is not convincing. Figure 4C has no loading control, such as probing for another protein like actin that shouldn't change in amount from well to well (loading equal protein levels is standard but not sufficient). Furthermore, the quantification of the band intensities does not match the image of the bands in that there isn't an almost 2.5-fold increase in the Fgf1 band under hypoxic conditions (16 to 41). Nor does the graph in 4C (right) look like there is a 2-fold change. For example, figure EV 2 shows a 1.24-fold change that is striking so a 2.5-fold increase should very apparent. Overall, this figure was not clearly explained other than it is a western, thus it is not clear what the graph is or how to interpret this panel.

The hairpin control for the negative control for IRES activity is better than no control but it is not sufficient to rule out other artifacts such as cryptic promoter activity. One thing that would be very informative is to know whether the first and second cistron in the dicistronic reporter are in the same reading frame or not. Also, if the hairpin control has the ORFs in the same reading frame. The impression this reviewer gets is that the control is not really identical to the other IRES constructs, but a negative control used for other IRES constructs. Regardless, the hairpin IRES negative control is not explained in the methods. A true negative control would be the same reporter with an insert that didn't have IRES activity.

Nuclearized – used in Figure 6 legend and in the results should be changed to "relocalization to the nucleus" as nuclearized means to supply with nuclear weapons or deploy nuclear weapons in (such as outer space).

The description of "waves" of IRESs is not justified by the data as they did not clearly show that there are multiple IRESs that are activated at distinctly different times. It wasn't clear that there were 2 distinct classes of IRESs that were activated at 4 hours of hypoxia and a different class at 24 hours of hypoxia. Were there examples of IRES that were active at different times? Yes, but EMCV was up at 4 hours and 24 hours if their statistics are correct (Figure 4D). Also, the error bars were pretty large and the differences were not very big.

Reviewer #3:

The manuscript has improved with the new additions. The methods, raw data and corrections are more detailed, and this allows for a better understanding of the manuscript. In addition, the conclusions and Discussion section have been toned down. However, there are still numerous over-statements and several controls missing (some, by the way, said to be included).

Most importantly, the main conclusion of the paper, that VASH1 is an ITAF, is challenged by the lack of data showing that the protein indeed binds to endogenous IRES-containing mRNAs. The authors show that endogenous VASH1 binds to biotinylated IRESs, and that VASH1 affects the expression of one IRES reporter (that of FGF1), but no data are shown on the effect of VASH1 on endogenous transcripts or products. The effect on IRES reporters other than FGF1 is negligible, even if statistically significant for some. So VASH1 may very well be a functional ITAF for just one of the analyzed transcripts, without a generalized function in IRES-mediated translation or hypoxia.

Essential revisions:

1) For the reason stated above, I think the Title and Abstract are over-stated.

2) The authors conclude that most genes are not induced at the transcriptome level in hypoxic cardiomyocytes, while they are controlled at the polysome level. But the thresholds to consider induction are different in Figure 1 (transcriptomics) and Figure 2C (translatome).

3) Regarding Figure 1, an unbiased clustering to detect coordinated behavior of transcripts along hypoxia would be a good addition to the manuscript.

4) Figure 4C: The authors state that "IRES induction correlates with an increased expression of FGF1 protein". However, rather than an increase I see very similar bands and quantification peaks in this figure.

5) Figure 4D: The authors keep talking about activation "waves", when there is only one construct corresponding to a cellular mRNA (c-myc) that is activated at 24 hours, and where only a few IRESs (which the authors sustain represent all IRES-containing mRNAs involved in angiogenesis) have been analyzed altogether. This is another over-statement.

6) Figure 5: It is now clear that VASH1 binds RNA with high affinity, but the specificity is an issue. in vitro, the protein binds to all tested IRESs, and in the BIA-MS experiments VASH1 was detected 0/5 times bound to VEGFA, 2/5 times bound to FGF1, and 4/5 times bound to EMCV IRESs. The BIA-MS experiments were done in different conditions and were repeated just once per condition. From these data, the authors conclude that VASH1 shows specificity for FGF1 and EMCV in cellulo. In my view, additional assays are required to conclude that VASH1 shows specificity in cellulo, such as RIP-qPCR using oligos for endogenous mRNAs.

7) Figure 6: The authors state that "VASH1 immunodetection confirmed a strong expression of VASH1 at 4 hours of hypoxia, despite the decrease of its mRNA". They have not included these data in the manuscript. It is important that a Western blot showing the induction of the protein at 4hours of hypoxia compared to normoxia is included to back-up the polysomal RNA level analysis of part B.

Also in this figure, please use a term other than "nuclearized" to indicate an increase in size of VASH1 foci. This is a confusing term, as it implies the nuclear compartment.

8) Figure 7 shows that depletion of VASH1 affects FGF1 IRES activity under hypoxia. The effect on all other IRESs is negligible even if in some cases statistically significant. Therefore, I find the sentence "These data showed that VASH1 behaves as an activator ITAF in hypoxia, limited to FGF1, VEGFD and EMCV IRESs, while it has an inhibitory role on the activities of these IRESs in normoxia" a strong overstatement.

https://doi.org/10.7554/eLife.50094.sa1

Author response

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

Reviewer #1:

The manuscript by Prats et al., aims to better understand the role of IRESs in cardiomyocytes under hypoxic conditions. They show transcriptomic and polysome analysis that a lot of mRNAs that encode for proteins important for angiogenesis and lymph-angiogenesis are translationally regulated. They go on to show that many of these mRNAs have IRES activity, which is temporally regulated during hypoxia. Then they identify one ITAF, VASH1, that is required for FGF1 IRES activity. Overall, this is a very nice study that makes a significant contribution to our understanding of hypoxia in immortalized muscle cells. However, the biggest issue with this manuscript is the lack of details. Many of the experiments are not explained well enough in the methods or the figure legend for the reader to be confident that they are interpreting the results correctly.

We fully agree about the lack of details and we have tried to answer this concern the best we can do. We have added explanations in the Material and methods section and in the figure legends. The details of our modifications are provided below.

Essential revisions:

It would be useful for the authors to justify why they chose 4, 8 and 24 hours for their time course. It would be nice to at least see how the IRES activity changed over a more detailed time course from 30 minutes to 24 hours having 10-15 time points. This would provide a better picture of the timing of the IRES regulation. For example, if hypoxia is translationally regulated why wouldn't we expect differences in 30 minutes or less? It would also be more compelling that there were differences in the timing of translation of these IRES containing mRNAs if there was more of a trend especially since the differences are so small (which doesn't mean that they aren't real). Also, given that the discussion focuses on how in many tumor cell lines that these IRESs are only activated after 24 hours of hypoxia, it would be informative to see if time points later than 24 hours resulted in a more robust activation of IRES activity. This would be necessary to support one of their major conclusions that the timing of IRES-mediated translation is regulated during hypoxia.

We chose 4, 8 and 24 hours because usually hypoxia experiments are performed at 24 hours and we wanted to look at early hypoxia. In addition, expression of VEGFA mRNA (HIF-1 target) is highly increased after 8 hours (Figure 1 and EV Figure 3). At the translational level we have previously published that the FGF2 IRES is activated after 4 hours of hypoxia in another cell type (Conte et al., 2008). However, we agree that IRES-dependent translation could occur earlier. As suggested by the reviewer we performed the time course for the FGF1 IRES and finally found that the peak of activity (never tested before) is at 6 hours. This kinetics has been added in Figure 4A. Interestingly, the ratio LucF/LucR decreases at 1 hour, before starting to increase. We could not go after 24 hours as the cells do not resist.

Figure 2A the authors claim that "The polysome profile displayed, as expected, a strong decrease of global translation (Figure 2A)." However, if there was a decrease in polysomes then there should be a corresponding increase in free subunits (40S and 60S), which is not seen. Rather their data show poor translational efficiency during normoxia and the reduced polysome peaks appear to be due to loading of less lysate. With different levels of lysate on the gradient the only definitive way to show that there is a significant reduction is to calculate the polysome (2-mer and up) to monosome (80S peak) ratio. The P/M should be shown for both normoxia and hypoxia polysomes.

We agree that the polysome profile provided in the first version did not show any decrease of the P/M ratio and that less lysate had been loaded. We have reproduced the experiment, and a correct polysome profile is now shown, with a P/M ratio that decreases from 1.55 to 1.40. It was calculated as suggested by the reviewer, polysome (disome and up) to monosome. The translational efficiency is indeed poor in normoxia in these cells (often observed with non cancerous cells), but still decreases in hypoxia.

In Figure 2 it is not clear how they calculated the fold change of the polysome associated mRNA under hypoxia. This is not adequately described in the figure legend and is not in the methods at all under Polysomal RNA preparation. Which fractions were collected for polysomes? Which fractions were free RNA? Whether all the fractions in the gradient were measured, which is important since equivalent levels of lysate was not used. How the statistics were calculated. The specific statistical test used should be in the figure legend rather than listing of all tests used in the manuscript in the Materials and methods section. What is meant by "normalized to normoxia as in Figure 1" (which isn't explained in Figure 1 either).

Indeed we did not explain enough how we calculated the polysomal mRNA fold change: the RQ (Relative quantification) was measured by the 2-DDCT method by calculating, for each mRNA, first the ratio of polysome-bound (disome fractions and up, fractions 19-27) to total RNA (from cell lysates before gradient loading), then the ratio hypoxia to normoxia which provided the fold change. When inferior to 1 (meaning a decrease), the RQ was expressed as -1/RQ to show the fold decrease. Then, the ratio polysome bound/total RNA was calculated to express the fold change in polysome recruitment. For more clarity, we have now explained this in Figure 2 legend and added all values in EV Table 2. The statistical test used in the manuscript is the Student test except for Figure 6 where we used ANOVA. We have added this information in each figure legend.

In Figure 3 it is not clear again how the Fold-change in polysomal mRNA was calculated. Comparing it to polysome associated in normoxia? Or looking at the shift from monosomes to polysomes under hypoxic conditions. The first has issues with sample differences and these would need to be adequately addressed. The latter approach must be clearly detailed. Alternatively, if only the polysomes were examined then this would have issues with sample variation since generally polysomes are internally controlled and if only a fraction of the polysomes are assayed then it becomes impossible to know what fraction of the mRNA was polysome associated.

Figure 3 just presents an extract from the data of Figure 1 and Figure 2 corresponding to the IRES-containing mRNAs. All the detailed values are in EV Table 1 for Figure 3A and in EV Table 2 for Figure 3A. The mode of calculation is also indicated in Figure 1 and Figure 2 legends and in EV Table 1 and Table 2 legends. For each mRNA, first the ratio of polysome-bound (disome fractions and up, pooled fractions 19-27) to total RNA (from cell lysates before gradient loading) was calculated, then the ratio hypoxia to normoxia, which provided the fold change. When inferior to 1 (meaning a decrease), the RQ was expressed as -1/RQ to show the fold decrease.

In Figure 4 they report that representative assays are shown. First the raw data needs to be reported in the supplementary material for all the replicates. This would include an empty vector with no IRES to show the background levels. Second, it is not clear why they are reporting representative data since it is a ratio of Fluc/Rluc which in theory should correct for differences in transduction efficiency. There is a significant concern that the differences observed are due to changes in Rluc values going down during hypoxia rather than a real increase in Fluc values. This should be cleared up by showing the raw data.

A table with the raw data (EV Table 3) has been provided for three independent experiments each including three biological replicates (n=9 for each IRES). An empty lentivector with an hairpin in place of an IRES (previously used in Morfoisse et al., 2014 by example) has been tested and provides LucF/LucR ratio lower than with all IRES and no significant difference betqeen normoxia and hypoxia)(EV Table 3J). The new Figure 4 shows the means of the 9 values). You can see in EV Table 3 that there is no decrease of LucR values at 4 and 8 hours, and that it only starts to decrease at 24 hours (and even not in all the experiments).

Table 2 the legend says "Relative quantification (RQ) of gene expression in hypoxia was calculated using the 2-ΔΔCT method (polysomal RNA/free RNA normalized to normoxia)." Does this mean that the data was normalized to normoxia either with or without siRNA KD of VASH1? Or was the KD data normalized to normoxia with VASH1 present in the cells? This should be clarified

EV table 2 has been modified for more clarity. The ΔCT values are presented for polysome bound and total RNA, as well as the RQ of bound/total ratio in normoxia and in hypoxia. Then the ratio of hypoxia/normoxia is presented, providing the fold change. We haved added the values for a second Fluidigm PCR array at 4 hours, and a PCR array at 24 hours. In each array, each gene was measured in triplicates. As regards the Fluidigm values with the siVASH1, we decided to remove them from the paper because they were not satisfactorily reproduced and thus, we could not really draw conclusions from them.

The argument that the authors make for mRNAs that are translationally up-regulated by the VASH1 ITAF need VASH1 to re-localize to the nucleus to form the "IRESome". However, there is no data to show that preventing re-localization of VASH1 to the nucleus affects FGF1 IRES activity. It seems just as likely that the nuclear localization of the VASH1 may be for another function. Also, given the increase in VASH1 protein under hypoxia and its translational up-regulation, its overall nuclear to cytoplasmic distribution may remain unchanged but is more apparent in the nucleus at the higher expression levels. There is clearly VASH1 in the nucleus under normoxic conditions.

Indeed, we admit that have no argument here showing that VASH1 has to be relocalized in the nucleus in hypoxia. In the revised version of Figure 6, we obtained higher quality pictures and performed quantification of the VASH1 positive foci in nucleus and in cytosol. Indeed, we clearly see VASH1 in the nucleus in normoxia and we do not measure any significant variation of the foci number between hypoxia and normoxia. However, the foci size clearly increases upon hypoxia. This suggests that newly synthesized VASH1 (shown by VASH1 mRNA recruitment in polysome Figure 6B) would localize in stress granules, as previously shown for other ITAFs such as hnRNPA1. Increase of stress granule size upon stress has been reported in a recent paper by Moon et al., 2019. As it is too early to show a model with the present data, we have removed the model proposed in the first version of Figure 7.

The authors should address why their reporter data does not match with their polysome data in that the reporter data shows the VEGFAa and VEGFAb IRESs were not increased above normoxic levels at 4 hours hypoxia (Figure 4C) whereas the polysome data shows they clearly are translationally upregulated (Figure 2 and Figure 3B). It is possible that they would see more dramatic changes in IRES activity of the reporters if they used a promoter such as the SV40 promoter, which is expressed at more physiological levels. There is some concern that using the higher expressing CMV promoter for these cellular IRESs reporters could saturate ITAFs therefore muting differences in IRES translation under hypoxic conditions for their reporters.

Indeed, we have no clear explanation for this difference. In the table with have never any activation at 4 hours for these two IRESs. The difference may come from a difference between the reporter and the endogenous mRNAs, or from a delay between the recruitment in polysomes and the luciferase accumulation. As we cannot explain this difference between 4 hours and 8 hours, we have modified our conclusion and we mention “early” hypoxia”, which includes 4 hours and 8 hours. We agree it would have been interesting to use with another promoter, as we have previously published ourselves that the promoter may affect IRES activity (Conte et al., 2009, Ainaoui et al., 2015), but it would have been a huge work to start again all these experiments (make all the constructs and lentivector production with another promoter, and perform all these experiments that are not easy with cardiomyocytes), whereas we observe significant data with the present promoter.

The methods for the real-time qPCR experiments is not clear. See the MIQE publication for best practices. We don't know how many replicates were measured, whether there were multiple dilutions of the sample measured, what the efficiency for the reactions were, which becomes critical given the small differences that were reported between samples. See "The MIQE Guidelines: Minimum Information for Publication of Quantitative Real-Time PCR Experiments".

We have added details in the Material and methods section about the replicates and dilutions, according to the MIQE guidelines.

Reviewer #2:

Hantelys and colleagues analyse changes in transcriptome and polysome association of mRNAs encoding lymphangiogenic factors during hypoxia. They observe differential activation of IRES-dependent translation along the hypoxic response and propose VASH1 as a protein binding and regulating the activity of certain IRESes. However, the manuscript at its present point is not convincing. There are several important controls missing in this study which are needed to support the conclusions:

1) Figure 2A: A decrease in polysomes should be accompanied by an increase in sub-polysomes or monosomal fractions. This does not seem to be the case of the profile shown here, suggesting that less material has been loaded in the gradient.

As mentioned above for reviewer 1, we agree that the polysome profile provided in the first version did not show any decrease of the P/M ratio and that less lysate had been loaded. We have reproduced the experiment, and a correct polysome profile is now shown, with a P/M ratio that decreases from 1.55 to 1.40. The translational efficiency is indeed poor in normoxia in these cells (often observed with non cancerous cells), but still decreases in hypoxia. Figure 2A has been modified and the new data have been included.

2) Table 3:

- What does the score mean and what would be a "good score"?

The score reflects the probability for a peptide to be present. The higher the score, the lower the probability of false positive: by example a score of 20 corresponds to a 5% probability of false positive. Specifically, a peptide is considered as really present if the score is above 12-13%. A sentence has been added in the Materials and methods section.

- Most proteins in this Table are detected with just one peptide. Proteins that are usual contaminants in affinity chromatography experiments (e.g. serum albumin) are also detected with one peptide here. How are true interactions distinguished from background?

We would prefer having detected several peptides. However, the score of 28% allows to conclude that this peptide is present. In addition, it has been detected in several samples for FGF1 and EMCV IRESs.

- From the data in Table 3, the choice of VASH1 as a potential ITAF seems aleatory because of the low level of detection and the fact that there is no difference between 4 hours (where the FGF1 IRES is active) and 8 hours (where the IRES is inactive). This protein is detected with only one peptide in all instances, so the "selective binding depending on the IRES and stress condition" alluded to by the authors in the main text is difficult to believe.

The peak of IRES activity is between 4 hours and 8 hours but can move within this window depending on the cell batch and the differentiation state of the cells. HL-1 cells are very sensitive to early hypoxia when they are beating but it is difficult to maintain them in this beating state. In the kinetics performed more recently with a new cell batch, the peak is around 6 hours of hypoxia. In the revised version we consider only early hypoxia (4 hours to 8 hours) versus late hypoxia (24 hours).

Another explanation is that the BIA-MS approach does not take into account all the cell parameters, in particular VASH1 intracellular localization. In addition, the complex with RNA is formed in vitro and not in cellulo. Indeed, there is no binding specificity in vitro but it may result of the absence of other IRESome partners. We have removed sentences about selective binding. However, we can clearly conclude about the RNA binding feature of VASH1 (Figure 5D-F).

Only one VASH1 peptide was identified, but in several samples (FGF1 4 hours and 8 hours of hypoxia, EMCV normoxia and hypoxia) and with a significant score. As mentioned above, the higher the score, the lower the probability of false positive: by example a score of 20 corresponds to a 5% probability of false positive.

VASH1 has been logically selected among the different bound proteins because of its involvement in angiogenesis and in stress tolerance (Sato and Atheroscler Thromb, 2015). It seemed apparent to us that VASH1, as a physiolocally relevant candidate, was the best choice to start the ITAF study. However, the other candidates remain interesting for the continuation of the project.

3) Figure 5: The RNA binding constant of VASH1 for FGF1 IRES is somewhat high and the RBDs are only predicted. Both RNA binding and the assignment of VASH RBDs should be backed-up by controls. For example, how does binding to VEGF IRES looks like? Is binding to FGF1 IRES competed by FGF1 and EMCV IRESes but not by VEGF IRES? Does mutation of the predicted RBDs in VASH1 eliminate binding? Importantly, does VASH1 bind to endogenous FGF1 mRNA?

These first binding experiments were done with the commercial VASH1 that contains only the 27kDa N-Terminal part of the protein coupled to GST, lacking the main predicted RNA binding C-terminal, domain (see Figure 5 and EV Figure 4). Since last year, we have started a collaboration with Yasufumi Sato (Sendaï, Japan), who sent us the full-length recombinant protein. The experiments with the full-length protein have been done for FGF1, VEGFA and EMCV IRESs. The KDs are 6.5 nM, 8 nM and 9,6 nM, respectively, that reflects a 400 times stronger affinity for FGF1 IRES than with the previous incomplete protein (2.8 mM, with the only protein that was available on the market!!). It is considered that RRM domains of PTB, for example, binds target RNA with a μM affinity. Here we are at the nanomolar level, which is considered the norm for transcription factors binding to DNA. With these data we can affirm that VASH1 full length isoform is an RNA-binding protein with a high affinity for RNA. VASH1 binding to the three IRES indicates that it is not sufficient to activate IRES activity and probably involves other partners in the cellular context. At the moment we have no data about the binding of VASH1 to endogenous FGF1 RNA. Figure 5 has been changed consequently, as well as the text (subsection “Identification of IRES-bound proteins in hypoxic cardiomyocytes reveals vasohibin1 as a new RNA-binding protein”).

4) Figure 6: Please, show a Western blot with and without VASH1 depletion to ensure that VASH1 antibody only detects the expected protein(s). I do not see a perinuclear localization pattern, but rather a cytoplasmic pattern.

We performed capillary Western that quantitatively show the VASH1 knock-down, whose efficiency is 59% (Figure 7). Due to the stability of VASH1 protein (superior to 24 hours as measured in EV Figure 5 of the revised version), it difficult to obtain a more efficient knock down. New (better) images were obtained with a confocal microscope. VASH1 was quantified (new Figure 6). To do that, VASH1 was colored in green in the cytosol and in red in the nucleus, indicating that VASH1 is present in both compartments. In hypoxia picture, a perinuclear staining becomes visible in hypoxia. However, we prefer to conclude only by saying that VASH is both cytosolic and nuclear. Interestingly, VASH1 is localized in nuclear and cytosolic bodies whose number does not increase but whose size significantly increases in cytoplasm upon stress, suggesting that it could be localized in stress granules.

5) Figure 7C: Given that the activity of the various IRESes in the bicistronic assay is weak (2-fold maximum), it is important to ensure that the reduction after VASH1 depletion is significant and is not affected by the variability inherent to the assays. For this reason, rather than showing one representative triplicate experiment and use standard error of the mean to measure significance, it is important to use the full set of biological replicates and show standard deviation. The same could be applied to Figure 4B-D.

The difficulty to obtain efficient VASH1 depletion is probably responsible for the weakness of the observed effects. We have reproduced several times the experiments and present three independent experiments for each IRES with all the crude values in EV Table 5. A reproducible significant effect has been obtained for FGF1 and VEGFD IRESs as well as a small effect on EMCV IRES. Maybe with a better knock down we could have obtained stronger data. The new version of Figure set uses the full set of independent experiments and biological replicates (n=9) to calculate the means. EV Table 5 recapitulates all the luciferase ratio values and clearly shows the reproducibility of our results. This has been done also for Figure 4 (EV Table 3). For each IRES and for each conditions n=9.

Also regarding Figure 7C, how do the authors explain the increase in VEGFA IRES activity upon VASH1 depletion if VASH1 does not bind to VEGFA IRES? The suggested inhibitory effect of VASH1 is more than discussible.

We agree with the reviewer: our data do not allow to propose an inhibitory effect of VASH1. This point has been withdrawn from the text.

6) How does depletion of VASH1 affect the polysomal distribution of FGF1 and VEFGA mRNAs in hypoxia? Table 2 shows a comparison of polysomes in hypoxia (4 hours) with siVASH1 polysomes in normoxia, while the correct comparison would be siControl polysomes in hypoxia (4 hours) with siVASH1 polysomes in hypoxia (4 hours).

Indeed, we wished to present the data with the siVASH1 in hypoxia, but we were not successful with that experiment because many mRNAs were not detectable. Probably not enough material was used in that experiment. As we had no opportunity to reproduce this fluidigm experiment, we have withdrawn it from EV Table 2: we agree that showing the data only in normoxia does not provide very useful information.

7) The Discussion section is highly speculative. The existence of "waves" of IRES-dependent regulation cannot be supported by the analysis of just 2-4 IRESes representing about 5% of the full set of translationally activated mRNAs under hypoxia. Furthermore, the physiological relevance of the potential regulation of IRES-dependent translation by VASH1 cannot be inferred by the established roles of VASH1, because each factor regulates multiple targets by different mechanisms. In addition, whether nuclear localization of VASH1 has any relevance for the function described here is an open question. The inference on regulons is too far-fetched given that there is only one member of each proposed regulon.

We agree that our discussion was too speculative. Even though these waves may exist, we were technically limited with the variations of the HL-1 sensitivity to hypoxia that can vary in the early times from one experiment to another, due to small variations of their phenotype with passages, with confluence etc. The more they are confluent and beating, the more they are sensitive in early hypoxia but it was really difficult to clearly establish the time of activation. In the kinetics presented in this new version, we see that there is a peak at 6h for FGF1 IRES. Due to these difficulties we encountered we prefer to give up on the distinction between the two waves of 4 hours and 8 hours, and we only mention early hypoxia (between 4 hours and 8 hours) and late hypoxia 24 hours. In addition, we should have analyzed more IRESs to conclude about this. Also, we agree we have not enough IRESs regulated by VASH1 at the moment to say that there is a regulon. This has been modified in the text.

Reviewer #3:

Figure 2:

While the data show that there are certainly less "heavy" polysomes in the hypoxic cells it would be better if the areas under the curves were calculated so that these data can be presented additionally as a polysome:subpolysome ratio to provide a stronger indication of the degree of initiation changes. This would also allow the extent of variation between experimentation to be observed. In addition, the authors need to carry out Met labelling or use some other measure to show that global protein synthesis rates are decreasing under these conditions. This should be accompanied by western analysis to show changes in for example, 4EBPs, eIF2α.

Indeed, the polysome profile provided in the first version did not show a decrease of the P/M ratio and probably less lysate had been loaded. We have reproduced the experiment, and a correct polysome profile is now shown, with a P/M ratio (calculated from the areas under the curves, ratio of disome and up to monosome) that decreases from 1.55 to 1.40. The translational efficiency is indeed poor in normoxia in these cells (often observed with non cancerous cells), but still decreases in hypoxia. Western blot (capillary electrophoresis) have been performed to detect, 4EBP, phospho4EBP, eIF2a and phospho-eIF2a. Unexpectedly, 4EBP appears as only one band in these cells (probably hypophosphorylated) and under hypoxia there is a small increase of this protein but no change on the level of phosphorylation. In contrast we observe a strong increase in eIF2a phosphorylation. These immunodetections have been added in Figure 2 and EV Figure 2.

Figure 3

The data show fold-changes in polysome mRNA localisation and in addition changes in transcription. Since the transcription and translation data have been obtained using arrays and, it is important to show how these transcripts vary across the gradients in hypoxic cells to confirm these findings, for a couple of mRNAs. Is it possible to calculate changes in translational efficiency from these data taking into account the transcriptional variation? This information would be very useful.

The mode of calculation of mRNA in polysome was not clearly explained in the first version. The values in the first version already corresponded to the ratio of mRNA in polysome to total RNA. It is now clearly explained in the text (Mat et meth pages 17 bottom, legends of Figure 2 and EV Table 2). In the new EV Table 2, we provide all the values (total RNA, polysomal RNA and ratio). In addition, we show it for a second fluidigm experiment with two times of hypoxia (4 hours, 24 hours).

Figure 4

The data show how the IRES function changes, but it is important to show how these data correlate with changes in expression of the corresponding proteins. For example, is there an increase in VEGF of 1.9 fold after 8 hours of hypoxia? If the half lives of these proteins are too long, pulse IP may need to be used.

We have measured the FGF1 expression in normoxia and hypoxia by capillary electrophoresis. The normalized areas show an 2.5 fold increase of the FGF1 protein in hypoxia.

Figure 6.

It is essential to measure the half-life of VASH1. Another explanation for the protein remaining expressed, even though the mRNA is decreasing, is that it is very stable with a long half-life. Therefore, it would not be turned over sufficiently during the course of the experiment to see a real difference.

The half-life of VASH1 was measured and is presented in EV Figure 5, comparatively to a control (P21). VASH1 half-life is superior to 24 hours, which can explain the difficulty to have large differences by knock down. The Western blot (capillary electrophoresis) shows a knock-down of 59% after 48 hours of siRNA treatment). This result is presented in the new Figure 7.

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

[…] Essential revisions:

The three original reviewers of your paper agree that you performed many additional important experiments and that the paper has improved considerably. However, as you can see two reviewers argue that some key issues remain to be addressed before the paper can be recommended for publication in eLife. These concerns were discussed and agreed on by all reviewers. In particular, it is critical to show that protein expression of endogenous VASH1 changes as the paper claims. You need to carry out pulse labeling IP of VASH1 to convincingly show changes in synthesis. You also ought to provide evidence that VASH1 specifically binds to the endogenous FGF-1 IRES. Another important issue that was raised by reviewer #2 concerns the statistical analysis of the data. The reviews of the three referees are attached so that you can read the detailed comments.

Reviewer #2:

This is a revised manuscript from Hantleys et al., entitled "Vasohibin1, a new IRES trans-acting factor for induction of (lymph)angiogenic factors in early hypoxia". […] They also identify an ITAF, VASH1, that binds to IRESs in the nanomolar range and that FGF1 IRES activity depends on this ITAF during hypoxia.

However, there are a few things that reduce enthusiasm such as the Fgf1 western which doesn't look like there is an increase in protein levels under hypoxic conditions.

We performed other capillary Western experiments that show a convincing increase of FGF1 protein expression upon hypoxia. A new picture has been added in Figure 4C.

Also, the IRES activities increase during hypoxia, but only by 1.2 to 1.7-fold. Is this sufficient enough for a biological response? Perhaps this is why the western didn't show a significant change in FGF1 protein levels or perhaps the increase in IRES activity is muted simply due to using immortalized cells or a promoter that overexpresses the dicistronic RNA. It is difficult to say. Even the ITAF, VASH1, which they reported bound to EMCV and FGF1 IRESs but not VEGFA IRES still had similar binding Kd for all of the IRESs. Overall, many of the conclusions are based on small effects or minor discrepancies that weaken the main conclusions, however, it is possible that these small changes in IRES activity are sufficient to alter cellular programs.

Indeed, IRES activities do not increase very strongly upon hypoxia but these increases are significant. Is this sufficient enough for a biological response? The answer is yes if one refers to several previous publications. A paragraph about this issue has been included Discussion section (subsection “Pathophysiological impact of a moderate stimulation of translation”):

“A striking feature of our data is that the stimulation of IRES activities by hypoxia in cardiomyocytes is moderate, only by 1.3 to 1.7 fold. […] Globally, cellular IRESs show lower degree of activation than viral IRESs, as illustrated by Braunstein et al. who report that the HIF1 IRES is stimulated by 1.6 fold during hypoxia, while the VEGFA IRES is stimulated by 2 fold and the EMCV IRESs by 3.5 fold (29).”

The similar Kd of VASH1 for all IRESs indeed means that VASH1 does not bind specifically to a given IRES.

We have discussed the absence of specificity of VASH1 binding in the Discussion section (subsection “VASH1 impact on translational control can be positive or negative”). The data with the fluidigm PCR array show that VASH1 has a wide impact on recruitment into polysomes of more than 60% of the mRNAs detected in the array. This supports a wide binding of VASH1 to RNA. In addition, our hypothesis is that the IRESome is not limited to the interaction of a single protein with RNA, but is multi-partner, explaining that a given ITAF can be either an inhibitor or an activator. This has been shown in several previous reports for at least 10 other ITAFs (we made an update about this in our review article by Godet et al., 2019).

“Among the IRESs analyzed in the present study, the FGF1 IRES is the only one to be strongly regulated by VASH1 in hypoxia. However, VASH1 was also bound to the EMCV IRES in the BIA-MS experiment, and calculation of affinity constants does not reveal significant differences of affinity for FGF1, VEGFA or EMCV IRES. This apparent inconsistency finds an explanation if one considers the effect of VASH1 in normoxia: there is a trend of IRESs to be activated upon VASH1 depletion, significant for VEGFD and EMCV IRESs. Such data suggest that VASH1 binding is probably not specific to a given IRES, but that different VASH1 partners would be recruited in the IRESome and result in positive or negative effects of this ITAF. The hypothesis of a dual role for VASH1 in translational control is confirmed by the effect of VASH1 depletion on translatome: recruitment into polysomes is affected negatively or positively for 60-70% of genes both in normoxia and in hypoxia. Although the RNA binding ability of VASH1 has been clearly shown in the present study, we cannot affirm that VASH1 impact is direct for all of these mRNAs. Nevertheless, a dual role of activator and inhibitor has been reported for more than ten other ITAFs. Our hypothesis thus remains that the key of the regulation of IRES activity by ITAFs is not RNA binding specificity but rather IRESome multi-partner composition (10).”

The authors reference a Figure 3C when there isn't a figure 3C.

Indeed, it has been corrected.

One of the key figures in the paper that supports a major conclusion that IRES activity leads to increased Fgf1 protein levels is not convincing. Figure 4C has no loading control, such as probing for another protein like actin that shouldn't change in amount from well to well (loading equal protein levels is standard but not sufficient). Furthermore, the quantification of the band intensities does not match the image of the bands in that there isn't an almost 2.5-fold increase in the Fgf1 band under hypoxic conditions (16 to 41). Nor does the graph in 4C (right) look like there is a 2-fold change. For example, figure EV 2 shows a 1.24-fold change that is striking so a 2.5-fold increase should very apparent. Overall, this figure was not clearly explained other than it is a western, thus it is not clear what the graph is or how to interpret this panel.

All the “Western” experiments shown in this paper are capillary Western. Each sample is loaded on an individual capillary. The “Jess” device calculates the ratio of a peak of a given protein to the total proteins effectively loaded in the capillary. This is more accurate and quantitative than in the classical western. Additional explanations have been included in figure legends and in the Materials and methods section to explain this. In the following link you can find explanations about Jess: https://www.proteinsimple.com/simple_western_videos.html

The quantification of the bands do not match with the image because the quantification is normalized to the total proteins. However, we agree that the previous image (before normalization) was not convincing for the reader, thus the Jess Simple Western with FGF1 antibody has been run again and the resulting image clearly shows the increase of FGF1 protein.

The hairpin control for the negative control for IRES activity is better than no control but it is not sufficient to rule out other artifacts such as cryptic promoter activity.

The presence of a cryptic promoter has been checked and figures about that included in all our previous papers with these IRESs. The only 5’ UTR to show a promoter activity is the VEGFA, but this promoter has been mapped between the two IRESs (Bornes et al., 2007). See Figure 2 of this paper. The level of each cistron is quantified by RTqPCR and it is clearly visible when there is an internal promoter.

In addition to all the published data, we checked this issue by RT qPCR of LucR and LucF in transduced HL-1 cells. The data have not included in the paper but are shown in Author response image 1 for the FGF1 IRES. These data measure the ΔΔCT compared to EMCV IRES. The ratio of 1 indicated that there is the same amount of the two cistrons, thus no cryptic promoter.

Author response image 1
Download asset Open asset

One thing that would be very informative is to know whether the first and second cistron in the dicistronic reporter are in the same reading frame or not. Also, if the hairpin control has the ORFs in the same reading frame. The impression this reviewer gets is that the control is not really identical to the other IRES constructs, but a negative control used for other IRES constructs. Regardless, the hairpin IRES negative control is not explained in the methods. A true negative control would be the same reporter with an insert that didn't have IRES activity.

The bicistronic vector with a hairpin has been fully validated as a negative control in our previous publications (the first one is Creancier et al., 2000). We do not believe that this control can be questioned. We have looked to the ORF of the two cistrons. Indeed, the intergenic region is a multiple of 3 thus the two ORFs are in the same frame. This could generate reinitiation but apparently, we do not observe it. The intergenic region contains two stop codons in the frame. The hairpin is a very stable structure expected to block any reinitiation. Construction of the hairpin vector has been explained in the Materials and methods section and the intergenic sequence has been added in Supplementary file 8A. The palindrome is indicated in color.

taaACTAGACGCGCTCTCCGTGAACTAGCGTAGCTGACCGATATCGGTCAGCTACGCTAGTTCACGGAGAGCGCGACTAGTGGATCCatg

Nuclearized – used in Figure 6 legend and in the results should be changed to "relocalization to the nucleus" as nuclearized means to supply with nuclear weapons or deploy nuclear weapons in (such as outer space).

We are sorry for this mistake. It has been corrected.

The description of "waves" of IRESs is not justified by the data as they did not clearly show that there are multiple IRESs that are activated at distinctly different times. It wasn't clear that there were 2 distinct classes of IRESs that were activated at 4 hours of hypoxia and a different class at 24 hours of hypoxia. Were there examples of IRES that were active at different times? Yes, but EMCV was up at 4 hours and 24 hours if their statistics are correct (Figure 4D). Also, the error bars were pretty large and the differences were not very big.

The term of waves has been removed from the text. With the new statistical analysis, we see that EMCV and c-myc are not induced in early hypoxia. Whereas the other ones are, except for VEGFA which is almost inactive if one looks at the values in Supplementary file 3. The error bars are large because this is standard deviation, and because we have regrouped a lot of independent experiments where the cells did not exhibit an identical sensitivity to hypoxia depending on various parameters. Despite of this, the Mann-Whitney test shows data with p-values between 0.05 and 0.0001.

Reviewer #3:

[…] Most importantly, the main conclusion of the paper, that VASH1 is an ITAF, is challenged by the lack of data showing that the protein indeed binds to endogenous IRES-containing mRNAs. The authors show that endogenous VASH1 binds to biotinylated IRESs, and that VASH1 affects the expression of one IRES reporter (that of FGF1), but no data are shown on the effect of VASH1 on endogenous transcripts or products. The effect on IRES reporters other than FGF1 is negligible, even if statistically significant for some. So VASH1 may very well be a functional ITAF for just one of the analyzed transcripts, without a generalized function in IRES-mediated translation or hypoxia.

Due to the VASH1 antibody which did not work in IP, it was not possible to obtain the RIP data despite several attempts. However, we address the issue of VASH1 effect on endogenous mRNAs by analyzing the effect of VASH1 depletion on translatome by fluidigm deltagenes PCR array: there is showing a strong effect on the translatome with 60-70% of mRNAs whose recruitment into polysomes varies positively or negatively (Figure 8, Supplementary file 7). This wide functional impact of VASH1, although not meaning that the interaction is direct (but RIP does not any more prove a direct interaction), strongly suggests that VASH1 interacts with numerous transcripts, directly or indirectly. The direct binding of VASH1 to RNA is clearly shown by the affinity constants measured in Figure 5.

Thus, paper shows that VASH1 is an ITAF only for FGF1 IRES, however we have preliminary data with other IRESs, for example IGF1R whose activity is sensitive to VASH1 depletion in hypoxia. However, we have not yet sufficiently reproduced this experiment (n=3) to publish it (F. David, unpublished). It is however consistent with the fluidigm data with siVASH1 (Supplementary file 7) where the IGF1R mRNA recruitment into polysomes decreases by 3.5 fold in hypoxia.

Essential revisions:

1) For the reason stated above, I think the Title and Abstract are over-stated.

We agree. The Title and the end of the Abstract have been changed.

2) The authors conclude that most genes are not induced at the transcriptome level in hypoxic cardiomyocytes, while they are controlled at the polysome level. But the thresholds to consider induction are different in Figure 1 (transcriptomics) and Figure 2C (translatome).

This has been corrected. The same threshold of 1.5, has been used for all the experiments. This new threshold is interesting as it allows to see differences between the different times of transcription. Even with this threshold (the previous threshold was 1), polysome recruitment remains activated for 94% of the genes. The conclusions in the text have been slightly modified following the changes of percentages that mostly concern transcriptome.

3) Regarding Figure 1, an unbiased clustering to detect coordinated behavior of transcripts along hypoxia would be a good addition to the manuscript.

We agree with this comment, but it was technically difficult for us to add this.

4) Figure 4C: The authors state that "IRES induction correlates with an increased expression of FGF1 protein". However, rather than an increase I see very similar bands and quantification peaks in this figure.

Quantification of the bands does not match with the image because it is normalized to the total proteins, that are not represented on the “numerical” blot. However, we agree that the previous image (before normalization) was not convincing, thus the Jess (capillary) Simple Western with FGF1 antibody has been run again and the resulting image clearly shows the increase of FGF1 protein.

5) Figure 4D: The authors keep talking about activation "waves", when there is only one construct corresponding to a cellular mRNA (c-myc) that is activated at 24 hours, and where only a few IRESs (which the authors sustain represent all IRES-containing mRNAs involved in angiogenesis) have been analyzed altogether. This is another over-statement.

We agree. The term of “waves” has been removed from the text. However, our data clearly show that induction of IRES activity for (lymph)angiogenic factor mRNAs occurs in early hypoxia. The new statistical test (Mann-Whitney) confirm that it occurs, except for VEGFA IRES a (which has a very poor IRES activity anyway in these cells).

6) Figure 5: It is now clear that VASH1 binds RNA with high affinity, but the specificity is an issue. in vitro, the protein binds to all tested IRESs, and in the BIA-MS experiments VASH1 was detected 0/5 times bound to VEGFA, 2/5 times bound to FGF1, and 4/5 times bound to EMCV IRESs. The BIA-MS experiments were done in different conditions, and were repeated just once per condition. From these data, the authors conclude that VASH1 shows specificity for FGF1 and EMCV in cellulo. In my view, additional assays are required to conclude that VASH1 shows specificity in cellulo, such as RIP-qPCR using oligos for endogenous mRNAs.

Indeed, our data show (and we conclude with this) that VASH1 does not bind specifically to a given IRES. Due to the VASH1 antibody which did not work in IP, it was not possible to obtain the RIP data despite several attempts. However, the wide effect of VASH1 depletion on recruitment into polysomes of more than 60% of the mRNAs detected in the array does not support a specific binding of VASH1 to RNA. Our hypothesis is that the IRESome is not limited to the interaction of a single protein with RNA, but is multi-partner, explaining that a given ITAF can be either an inhibitor or an activator. Dual ITAF effect has been shown in several previous reports for at least 10 other ITAFs (we made an update about this in our review article by Godet et al., 2019). We discuss this issue in the Discussion section.

7) Figure 6: The authors state that "VASH1 immunodetection confirmed a strong expression of VASH1 at 4 hours of hypoxia, despite the decrease of its mRNA". They have not included these data in the manuscript. It is important that a Western blot showing the induction of the protein at 4hours of hypoxia compared to normoxia is included to back-up the polysomal RNA level analysis of part B.

Also in this figure, please use a term other than "nuclearized" to indicate an increase in size of VASH1 foci. This is a confusing term, as it implies the nuclear compartment.

We performed capillary Simple Western and clearly show by that VASH1 is induced at 4 hours. This has been included in Figure 6C. The complete kinetics is included below, with the quantification. We have removed “nuclearized”. Sorry again for this mistake.

Author response image 2
Download asset Open asset

8) Figure 7 shows that depletion of VASH1 affects FGF1 IRES activity under hypoxia. The effect on all other IRESs is negligible even if in some cases statistically significant. Therefore, I find the sentence "These data showed that VASH1 behaves as an activator ITAF in hypoxia, limited to FGF1, VEGFD and EMCV IRESs, while it has an inhibitory role on the activities of these IRESs in normoxia" a strong overstatement.

We have modified the text in subsection “Vasohibin1 is a new ITAF active in early hypoxia”: “In contrast, in hypoxia, VASH1 knock-down resulted in strong decrease of FGF1 IRES activity, by 64%, whereas it did not significantly affect the other IRESs (Figure 7D). These data showed that VASH1 behaves as an activator of FGF IRES in hypoxia, while it tends to inhibit several IRESs in normoxia (Figure 7C).”

https://doi.org/10.7554/eLife.50094.sa2

Article and author information

Author details

  1. Fransky Hantelys

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology
    Contributed equally with
    Anne-Claire Godet and Florian David
    Competing interests
    No competing interests declared

  2. Anne-Claire Godet

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology
    Contributed equally with
    Fransky Hantelys and Florian David
    Competing interests
    No competing interests declared

  3. Florian David

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Investigation, Methodology
    Contributed equally with
    Fransky Hantelys and Anne-Claire Godet
    Competing interests
    No competing interests declared

  4. Florence Tatin

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Formal analysis, Methodology
    Competing interests
    No competing interests declared

  5. Edith Renaud-Gabardos

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Resources, Formal analysis, Investigation
    Competing interests
    No competing interests declared

  6. Françoise Pujol

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  7. Leila H Diallo

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  8. Isabelle Ader

    UMR 1031-STROMALAB, Inserm, CNRS ERL5311, Etablissement Français du Sang-Occitanie (EFS), National Veterinary School of Toulouse (ENVT), Université de Toulouse, UPS, Toulouse, France
    Contribution
    Supervision, Investigation
    Competing interests
    No competing interests declared

  9. Laetitia Ligat

    UMR 1037-CRCT, Inserm, CNRS, Université de Toulouse, UPS, Pôle Technologique-Plateau Protéomique, Toulouse, France
    Contribution
    Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared

  10. Anthony K Henras

    UMR 5099-LBME, CBI, CNRS, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Supervision, Methodology
    Competing interests
    No competing interests declared

  11. Yasufumi Sato

    Department of Vascular Biology, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan
    Contribution
    Resources, Formal analysis, Methodology
    Competing interests
    No competing interests declared

  12. Angelo Parini

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Supervision, Funding acquisition
    Competing interests
    No competing interests declared

  13. Eric Lacazette

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Supervision, Methodology
    Competing interests
    No competing interests declared

  14. Barbara Garmy-Susini

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Supervision, Methodology
    Competing interests
    No competing interests declared

  15. Anne-Catherine Prats

    UMR 1048-I2MC, Inserm, Université de Toulouse, UPS, Toulouse, France
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Project administration
    For correspondence
    anne-catherine.prats@inserm.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5282-3776

Funding

Region Midi-Pyrenees

  • Anne-Catherine Prats

AFM-Téléthon

  • Edith Renaud-Gabardos
  • Anne-Catherine Prats

Association pour la Recherche sur le Cancer

  • Anne-Catherine Prats

European Commission (REFBIO VEMT)

  • Anne-Catherine Prats

Fondation Toulouse Cancer-Sante

  • Barbara Garmy-Susini

Agence Nationale de la Recherche (ANR-18-CE11-0020-RIBOCARD)

  • Anne-Catherine Prats

Ligue Contre le Cancer

  • Fransky Hantelys
  • Anne-Claire Godet

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Our thanks go to JJ Maoret and F Martins from the Inserm UMR1048 GeT-TQ plateau of the GeT platform Genotoul (Toulouse), F Lopez and L Tonini from the proteomic platform Genotoul (Toulouse), J Iacovoni from the Inserm UMR 1048 bioinformatics plateau, as well as L van den Berghe and C Segura from the Inserm UMR1037 vectorology plateau (Toulouse) and A Lucas from the We-Met Functional Biochemistry Facility (Toulouse). We also thank J Cavaille, C Müller and V Poinsot for helpful discussion and W Claycomb for providing HL-1 cells.

This work was supported by Région Midi-Pyrénées, Association Française contre les Myopathies (AFM-Téléthon), Association pour la Recherche sur le Cancer (ARC), European funding (REFBIO), Fondation Toulouse Cancer Santé and Agence Nationale de la Recherche ANR-18-CE11-0020-RIBOCARD. FH received fellowships from the Région Midi-Pyrénées and from the Ligue Nationale Contre le Cancer (LNCC). ERG had a fellowship from AFM-Telethon. AC Godet had a fellowship from LNCC.

Senior Editor

  1. James L Manley, Columbia University, United States

Reviewing Editor

  1. Nahum Sonenberg, McGill University, Canada

Version history

  1. Received: July 18, 2019
  2. Accepted: December 9, 2019
  3. Accepted Manuscript published: December 9, 2019 (version 1)
  4. Version of Record published: January 7, 2020 (version 2)

Copyright

© 2019, Hantelys et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,548
    Page views
  • 243
    Downloads
  • 13
    Citations

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

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Fransky Hantelys
  2. Anne-Claire Godet
  3. Florian David
  4. Florence Tatin
  5. Edith Renaud-Gabardos
  6. Françoise Pujol
  7. Leila H Diallo
  8. Isabelle Ader
  9. Laetitia Ligat
  10. Anthony K Henras
  11. Yasufumi Sato
  12. Angelo Parini
  13. Eric Lacazette
  14. Barbara Garmy-Susini
  15. Anne-Catherine Prats
(2019)
Vasohibin1, a new mouse cardiomyocyte IRES trans-acting factor that regulates translation in early hypoxia
eLife 8:e50094.
https://doi.org/10.7554/eLife.50094

Categories and tags

Research organism

Further reading

    1. Cell Biology
    2. Chromosomes and Gene Expression

    Long non-coding RNA Neat1 and paraspeckle components are translational regulators in hypoxia

    Anne-Claire Godet, Emilie Roussel ... Anne-Catherine Prats
    Research Article Updated

    Internal ribosome entry sites (IRESs) drive translation initiation during stress. In response to hypoxia, (lymph)angiogenic factors responsible for tissue revascularization in ischemic diseases are induced by the IRES-dependent mechanism. Here, we searched for IRES trans-acting factors (ITAFs) active in early hypoxia in mouse cardiomyocytes. Using knock-down and proteomics approaches, we show a link between a stressed-induced nuclear body, the paraspeckle, and IRES-dependent translation. Furthermore, smiFISH experiments demonstrate the recruitment of IRES-containing mRNA into paraspeckle during hypoxia. Our data reveal that the long non-coding RNA Neat1, an essential paraspeckle component, is a key translational regulator, active on IRESs of (lymph)angiogenic and cardioprotective factor mRNAs. In addition, paraspeckle proteins p54nrb and PSPC1 as well as nucleolin and RPS2, two p54nrb-interacting proteins identified by mass spectrometry, are ITAFs for IRES subgroups. Paraspeckle thus appears as a platform to recruit IRES-containing mRNAs and possibly host IRESome assembly. Polysome PCR array shows that Neat1 isoforms regulate IRES-dependent translation and, more widely, translation of mRNAs involved in stress response.

    1. Cell Biology

    Rab12 is a regulator of LRRK2 and its activation by damaged lysosomes

    Xiang Wang, Vitaliy V Bondar ... Anastasia G Henry
    Short Report Updated

    Leucine-rich repeat kinase 2 (LRRK2) variants associated with Parkinson’s disease (PD) and Crohn’s disease lead to increased phosphorylation of its Rab substrates. While it has been recently shown that perturbations in cellular homeostasis including lysosomal damage can increase LRRK2 activity and localization to lysosomes, the molecular mechanisms by which LRRK2 activity is regulated have remained poorly defined. We performed a targeted siRNA screen to identify regulators of LRRK2 activity and identified Rab12 as a novel modulator of LRRK2-dependent phosphorylation of one of its substrates, Rab10. Using a combination of imaging and immunopurification methods to isolate lysosomes, we demonstrated that Rab12 is actively recruited to damaged lysosomes and leads to a local and LRRK2-dependent increase in Rab10 phosphorylation. PD-linked variants, including LRRK2 R1441G and VPS35 D620N, lead to increased recruitment of LRRK2 to the lysosome and a local elevation in lysosomal levels of pT73 Rab10. Together, these data suggest a conserved mechanism by which Rab12, in response to damage or expression of PD-associated variants, facilitates the recruitment of LRRK2 and phosphorylation of its Rab substrate(s) at the lysosome.

    1. Cell Biology

    Genome-wide screen reveals Rab12 GTPase as a critical activator of Parkinson’s disease-linked LRRK2 kinase

    Herschel S Dhekne, Francesca Tonelli ... Suzanne R Pfeffer
    Research Advance Updated

    Activating mutations in the leucine-rich repeat kinase 2 (LRRK2) cause Parkinson’s disease. LRRK2 phosphorylates a subset of Rab GTPases, particularly Rab10 and Rab8A, and we showed previously that these phosphoRabs play an important role in LRRK2 membrane recruitment and activation (Vides et al., 2022). To learn more about LRRK2 pathway regulation, we carried out an unbiased, CRISPR-based genome-wide screen to identify modifiers of cellular phosphoRab10 levels. A flow cytometry assay was developed to detect changes in phosphoRab10 levels in pools of mouse NIH-3T3 cells harboring unique CRISPR guide sequences. Multiple negative and positive regulators were identified; surprisingly, knockout of the Rab12 gene was especially effective in decreasing phosphoRab10 levels in multiple cell types and knockout mouse tissues. Rab-driven increases in phosphoRab10 were specific for Rab12, LRRK2-dependent and PPM1H phosphatase-reversible, and did not require Rab12 phosphorylation; they were seen with wild type and pathogenic G2019S and R1441C LRRK2. As expected for a protein that regulates LRRK2 activity, Rab12 also influenced primary cilia formation. AlphaFold modeling revealed a novel Rab12 binding site in the LRRK2 Armadillo domain, and we show that residues predicted to be essential for Rab12 interaction at this site influence phosphoRab10 and phosphoRab12 levels in a manner distinct from Rab29 activation of LRRK2. Our data show that Rab12 binding to a new site in the LRRK2 Armadillo domain activates LRRK2 kinase for Rab phosphorylation and could serve as a new therapeutic target for a novel class of LRRK2 inhibitors that do not target the kinase domain.