Translational control of gene expression is an important regulator of adult stem cell quiescence, activation and self-renewal. In skeletal muscle, quiescent satellite cells maintain low levels of protein synthesis, mediated in part through the phosphorylation of eIF2α (P-eIF2α). Pharmacological inhibition of the eIF2α phosphatase with the small molecule sal003 maintains P-eIF2α and permits the expansion of satellite cells ex vivo. Paradoxically, P-eIF2α also increases the translation of specific mRNAs, which is mediated by P-eIF2α-dependent read-through of inhibitory upstream open reading frames (uORFs). Here, we ask whether P-eIF2α-dependent mRNA translation enables expansion of satellite cells. Using transcriptomic and proteomic analyses, we show a number of genes associated with the assembly of the spindle pole to be upregulated at the level of protein, without corresponding change in mRNA levels, in satellite cells expanded in the presence of sal003. We show that uORFs in the 5′ UTR of mRNA for the mitotic spindle stability gene Tacc3 direct P-eIF2α-dependent translation. Satellite cells deficient for TACC3 exhibit defects in expansion, self-renewal and regeneration of skeletal muscle.
Skeletal muscle regeneration relies on a population of resident adult stem cells, named ‘satellite cells’ for their position around the outside of myofibers and underneath the basal lamina (Mauro, 1961). Satellite cells express members of the paired homeodomain family of transcription factors PAX7 and, in a subset of muscle, PAX7 and PAX3 (Relaix et al., 2006). Satellite cells are mitotically quiescent and activate the myogenic program and the cell cycle in response to muscle injury. Activated satellite cells exhibit remarkable proliferative capacity to rapidly expand the population of myogenic progenitors required to efficiently regenerate muscle. Moreover, this proliferative phase is marked by symmetric cell divisions, which drive expansion of the satellite cell pool, and asymmetric cell divisions, mediated in part through an EGFR-AuroraA kinase signaling pathway, which ensure satellite cell self-renewal (Wang et al., 2019). How satellite cells achieve this proliferative capacity, while maintaining the fidelity of cell division, remains unclear.
Quiescent satellite cells have few activated mitochondria and generate low levels of ATP (Rocheteau et al., 2012; Rodgers et al., 2014). To manage available energy resources, quiescent satellite cells likely maintain low rates of protein synthesis, mediated in part by the phosphorylation of translation initiation factor eIF2α (Zismanov et al., 2016). Phosphorylation of the eIF2α subunit of the eIF2 complex turns eIF2 into a competitive inhibitor of the guanine nucleotide exchange factor eIF2B to prevent the recycling of the eIF2-GTP-initiator methionyl-tRNA ternary complex (eIF2-GTP-tRNAMet) needed to deliver the first amino acid to the ribosome (Krishnamoorthy et al., 2001). Genetic inactivation of eIF2α phosphorylation (P-eIF2α) in satellite cells leads to increased global protein synthesis, activation of the myogenic program and failure to self-renew. Pharmacological inhibition of eIF2α dephosphorylation with the small molecule sal003 (Boyce et al., 2005; Costa-Mattioli et al., 2007) maintains low levels of protein synthesis and enables ex vivo expansion of cultured satellite cells that retain their regenerative potential (Lean et al., 2019; Zismanov et al., 2016).
Although maintaining low levels of protein synthesis, quiescent satellite cells also repress the translation of specific transcripts that maintain satellite cells ‘primed’ to activate the myogenic program and the cell cycle. Myf5 transcripts are repressed by miR-31 and accumulate in cytoplasmic RNA granules (Crist et al., 2012), which require P-eIF2α for their assembly and maintenance (Zismanov et al., 2016). MyoD (Myod1) transcripts are repressed by the activity of RNA-binding proteins TTP (ZFP36; Hausburg et al., 2015) and STAUFEN 1 (STAU1; de Morrée et al., 2017), and transcripts required to activate the cell cycle, such as Dek, are repressed by miR-489 (Cheung et al., 2012).
Although the role of post-transcriptional mechanisms repressing gene expression in satellite cells is well documented, a more complete picture requires that we also understand how specific mRNAs escape repression and are translated efficiently. When cells are under stress, P-eIF2α-dependent mRNA translation is illustrated by selective translation of mRNAs for Atf4 and Chop (Ddit3), which are required to initiate an integrated stress response (ISR) (Palam et al., 2011; Vattem and Wek, 2004). These P-eIF2α-dependent mRNAs have inhibitory upstream open reading frames (uORFs) that prevent the translation of the main ORF under normal conditions, whereas stress-induced P-eIF2α permits the read-through of the uORFs to enhance the translation of the main ORF required for the ISR.
We propose a model by which P-eIF2α-dependent mRNA translation regulates satellite cell quiescence, self-renewal and expansion. In this study, we determined how satellite cells modify their transcriptome and proteome while expanded with sal003. We show that sal003 pushes satellite cell gene expression towards a progenitor cell phenotype, while preventing differentiation. By focusing on genes upregulated at the level of protein, without a corresponding increase in mRNA, we demonstrate that satellite cell expansion is mediated in part through P-eIF2α-dependent translation of mRNA for Tacc3 (transforming acidic coiled coil protein 3), an AuroraA kinase substrate that is essential for microtubule assembly, maintenance of the spindle pole and ensuring the fidelity of cell division (Burgess et al., 2015; Kinoshita et al., 2005).
Pharmacological inhibition of eIF2α dephosphorylation leads to expansion of satellite cells with distinct transcriptional profiles
To begin asking how sal003/P-eIF2α enables satellite cell expansion, we first used RNA-seq to determine global changes in gene expression in satellite cells isolated from diaphragm and abdominal muscle of adult Pax3GFP/+ mice (Montarras et al., 2005) after 4-day culture in the presence of 10 µM sal003 (Table S1; Zismanov et al., 2016). Principal component analysis (PCA) reveals a strong, global effect on the transcriptome, with sal003 treatment explaining 90% of the variance in the data (Fig. 1A). Consistent with the effect of sal003 to reduce global protein synthesis (Zismanov et al., 2016), we observed a general trend towards downregulation of genes that produced an asymmetrical volcano plot (Fig. 1B), as well as a larger number of downregulated genes (958 versus 383, P-value<0.00001). We generated a z-score heatmap to visualize selected genes that are upregulated or downregulated at least 2-fold in the presence of 10 µM sal003 (Fig. 1C).
To understand the mechanisms underlying changes in gene expression when satellite cells are cultured in the presence of sal003, we used Enrichr (Chen et al., 2013) to perform gene ontology (GO) analyses (GO Biological Process 2018), as well as to identify enrichment of consensus target genes for transcription factors (ENCODE and ChEA Consensus TFs from ChIP-X) and histone modifications (ENCODE Histone Modifications 2015). Amongst the 383 genes upregulated greater than 2-fold (log2>1), there is strong enrichment for GO Biological Process gene sets related to DNA replication, mitosis and cell cycle regulation (Fig. 1D). When satellite cells are cultured in the presence of sal003, we observe enrichment for target genes of known transcriptional regulators of myogenic progenitor survival and proliferation (E2f1, Sin3a, E2f6, E2f4, Foxm1) (Wang et al., 1996) (Fig. 1F). Upregulated genes also show enrichment for chromatin marks associated with actively transcribed genes in myogenic progenitors (H3K4me3, H3K9ac, H3K6me3) (Fig. 1H).
Amongst the 958 genes downregulated greater than 2-fold (log2<−1) there is strong enrichment for GO Biological Process gene sets related to differentiated skeletal muscle, including muscle contraction, muscle filament sliding, myofibril assembly and sarcomere organization (Fig. 1E). Amongst these genes, consensus target genes for transcription factor MYOD is predominant (Fig. 1G) and moreover, downregulated genes are enriched for chromatin marks common to repressed gene bodies (H3K27me3) in non-skeletal muscle tissue (Fig. 1I).
Pharmacological inhibition of eIF2α dephosphorylation modifies the satellite cell proteome
As sal003 inhibition of P-eIF2α dephosphorylation lowers rates of protein synthesis (Zismanov et al., 2016) and is expected to impact rates of mRNA translation independent of changes in transcription, we simultaneously asked how sal003 modifies the satellite cell proteome. We cultured satellite cells under normal conditions or in the presence of sal003, then labeled cell lysates with tandem mass tag (TMT) reagents to identify and quantify changes in protein expression by high resolution mass spectrometry (Fig. 2A; Table S2).
Of the 180 genes upregulated 2-fold or greater (log2>1) quantified by mass spectrometry, there is strong enrichment for genes expressed in mouse progenitor and stem cell lines (BioGPS, Fig. 2B). Upregulated genes are enriched in chromatid and chromosome separation (GO Biological Process, Fig. 2D), and are associated with the mitotic spindle, centromere and microtubules (GO Cellular Component, Fig. 2F). In contrast, the presence of 10 µM sal003 in satellite cell culture conditions downregulates 185 genes (2-fold, log2<−1) quantified by mass spectrometry. These genes are enriched in mouse skeletal muscle tissue (BioGPS, Fig. 2C), are associated with muscle contraction and muscle filament sliding (GO Biological Processes, Fig. 2E), and are components of striated muscle thin filament, actin cytoskeleton, focal adhesions and actomyosin (GO Cellular Component, Fig. 2G).
To reveal transcripts potentially translated in a P-eIF2α-dependent manner, we identified 140 genes that are upregulated at the level of protein (mass spectrometry, log2>1), without a corresponding increase reported by mRNA (RNA-seq, log 2<1) (Fig. 3A,B). Upregulated genes include inhibitors of myogenic differentiation (Pax7, Id3, Cabin1) and chromatin modifiers (Kdm5c, Kmt2a, Setd1a, Setd2) that may account for, in part, transcriptional changes in gene expression observed (Fig. 1). The most significantly represented class of genes are those involved in spindle assembly (Tacc3, Cdc20, Tpx2, Nedd1, Racgap1, Espl1, Spag5, Incenp, Vps4b, Kif2c, Ska3) (Fig. 3A,B; Table S3).
P-eIF2α-dependent translation of Tacc3 mRNA
We further focused our attention on Tacc3, which was the most significantly upregulated gene reported by mass spectrometry and is a representative gene of the most significantly enriched gene ontology (spindle pole). Tacc3 mRNA contains five uORFs in its 5′ untranslated region (UTR), potentially enabling selective translation of the main ORF for Tacc3 by P-eIF2α. TACC3 is a substrate of AuroraA kinase and functions at the centrosome to regulate microtubule nucleation, promote stability of the spindle apparatus and the fidelity of cell division (Burgess et al., 2015; Kinoshita et al., 2005). Inhibition of TACC3 by knockout, knockdown and pharmacological strategies reveal a role for TACC3 in maintaining or expanding adult and cancer stem cell populations, although it is dispensable for stem cell differentiation (Piekorz et al., 2002; Wurdak et al., 2010; Yao et al., 2016; Zhou et al., 2015).
First, we confirmed that 4-day culture of satellite cells in the presence of 10 µM sal003 increases TACC3 protein levels (Fig. 3C), without a corresponding change in Tacc3 mRNA (Fig. 3D), validating the identification of Tacc3 in our global gene expression profiles (Fig. 3A,B). Next, we cultured satellite cells in the presence of sal003 to induce P-eIF2α levels and in the presence of integrated stress response inhibitor (ISRIB), a small molecule inhibitor of P-eIF2α first identified using a high throughput cell-based screen for inhibitors of uORFs present in Atf4 mRNA. Mechanistically, ISRIB activates eIF2B to recycle the eIF2-GTP-tRNAMet ternary complex, bypassing the effect of P-eIF2α (Sidrauski et al., 2015). Enhanced eIF2B activity leads to translation initiation at the inhibitory uORFs. We showed that additional treatment of satellite cells with ISRIB lowers TACC3 protein levels to normal culture conditions, further linking Tacc3 mRNA translation to the effects of P-eIF2α (Fig. 3E). Finally, we used immunofluorescence with antibodies against PAX7 and TACC3, combined with EdU labeling, to reveal increased levels of TACC3 protein in proliferating PAX7(+) cells after 4-day culture in the presence of sal003 (Fig. 3F,G).
To ask whether uORFs present in the 5′ UTR of Tacc3 mRNA mediate P-eIF2α-dependent translation in a manner similar to Atf4, we cloned the 5′ UTRs of these two mRNAs upstream of a luciferase reporter. We generated additional control reporters that eliminate uORFs by mutation of the corresponding ATG start codon (Fig. 3H). We transfected 293 cells with Atf4 and Tacc3 P-eIF2α reporters, and further cultured 293 cells under normal conditions or in the presence of thapsigargin (TG) to induce high P-eIF2α levels (Vattem and Wek, 2004). P-eIF2α-dependent translation of the luciferase reporter occurs in the presence of TG when uORFs in Atf4 or Tacc3 are present, but not when these uORFs are eliminated by mutation of the start codon (Fig. 3H).
TACC3 is abundant in PAX7-expressing satellite cells, and downregulated upon differentiation
Next, we asked when TACC3 is expressed during the myogenic program. Low-level TACC3 expression is observed by immunoblotting cell lysates of fresh isolated satellite cells, compared with differentiating satellite cells that have been cultured for 4 days (Fig. 4A). To compare TACC3 expression in quiescent versus activated satellite cells, we first used magnetic cell sorting (MACS) to isolate satellite cells from uninjured and injured tibialis anterior muscle (TA) muscle. TACC3 is strongly upregulated in activated satellite cells in vivo, 3 days after cardiotoxin (ctx) injury (Fig. 4B,C). In activated satellite cells that remain associated with cultured single extensor digitorum longus (EDL) myofibers, strong immunolabeling of TACC3 is observed in an average of 56% of PAX7(+) satellite cells after 24 h in culture, and in 96% of satellite cells after 48 h in culture, in which it appears to be localized to perinuclear areas and to spindle poles (Fig. 4D,E).
During ex vivo culture of satellite cells, P-eIF2α is detectable in PAX7-expressing cells, but not in MYOD-expressing cells that have activated the myogenic program (Zismanov et al., 2016). We therefore asked whether TACC3 accumulates preferentially in PAX7-expressing cells. After 4-day culture of isolated satellite cells, TACC3 remains expressed in >80% of PAX7-expressing myogenic progenitors, in 65% of MYOD-expressing progenitors and in <20% of myogenin-expressing cells undergoing differentiation (Fig. 4F,G).
Tacc3 is required for expansion of self-renewing satellite cells ex vivo
As sal003 upregulates TACC3 and facilitates ex vivo expansion of self-renewing satellite cells, we next asked whether Tacc3 is required for satellite cell expansion and self-renewal. We examined satellite cells from Pax7CreERT2/+; Tacc3fl/fl mice (Murphy et al., 2011; Yao et al., 2007, 2016), such that tamoxifen (tmx) treatment would result in loss of TACC3 from PAX7-expressing satellite cells. First, satellite cells were isolated from tmx-treated Pax7CreERT2/+; Tacc3fl/fl mice and further cultured in the presence of 4-hydroxytamoxifen (4-OHT) to eliminate Tacc3 expression in cultured satellite cells (Fig. 5A-C). In the absence of TACC3, PAX7 is present diffusely within the nucleus and cytoplasm (Fig. 5B,D), and low PAX7 levels are confirmed by western blotting (Fig. 5E) and RT-qPCR (Fig. 5F). Satellite cells defective for Tacc3 have a pronounced defect to expand into large myocolonies ex vivo (Fig. 5G,H). Further analysis of myocolonies present after 4-day culture indicate precocious differentiation, as revealed by an increase in PAX7(−), MYOD(+) nuclei, and a decrease in self-renewal, as indicated by the near absence of PAX7(+), MYOD(−) nuclei representing the ‘reserve cell’ population (Fig. 5G,H). We used EdU incorporation assays to compare rates of proliferation between wild-type Pax7CreERT2/+ and Pax7CreERT2/+; Tacc3fl/fl satellite cells over the course of 4-day culture (Fig. 5I-K). We observe a marked decrease in EdU incorporation in the progeny of Pax7CreERT2/+; Tacc3fl/fl satellite cells at later time points in culture (days 3-4) (Fig. 5J), and moreover, the PAX7-expressing compartment of EdU(+) proliferating progeny is specifically reduced in Tacc3-deficient satellite cells (Fig. 5K). Next, we rescued TACC3 expression in satellite cells isolated from tmx-treated Pax7CreERT2/+; Tacc3fl/fl mice with lentiviral vectors driving Tacc3 expression under the phosphoglycerate kinase 1 (PGK) promoter (Fig. 5L). Increased TACC3 levels led to more robust PAX7 expression and localization of PAX7 to the nucleus (Fig. 5L-N).
We further asked whether TACC3 was important for the effect of sal003 in expanding satellite cells ex vivo. We isolated satellite cells from diaphragm and abdominal muscle from tmx-administered Pax7CreERT2/+ and Pax7CreERT2/+; Tacc3fl/fl mice and cultured them for 4 days under normal conditions or in the presence of 10 µM sal003. In the presence of sal003, wild-type Pax7CreERT2/+ satellite cells gave rise to >80% PAX7-expressing progeny after 4-day culture (Fig. S1A,C), as we have reported previously (Zismanov et al., 2016; Lean et al., 2019). In contrast, Pax7CreERT2/+; Tacc3fl/fl satellite cells failed to expand and instead gave rise to small colonies composed of fewer PAX7-expressing cells (Fig. S1B,C). Similar results were obtained for satellite cells isolated from hindlimb muscle (Fig. S1D-F).
The importance of TACC3 to promote satellite cell expansion, rates of proliferation and satellite cell self-renewal, while preventing precocious differentiation, is also confirmed by siRNA knockdown of Tacc3 (Fig. S2) and by TACC3 inhibition with the small compound KHS101 (Fig. S3;Wurdak et al., 2010).
Tacc3 deficient satellite cells expand poorly, leading to defects in muscle regeneration
To reveal a role for TACC3 in satellite cells in vivo, we inactivated Tacc3 by tmx administration to Pax7CreERT2/+; Tacc3fl/fl mice. In uninjured muscle, numbers of PAX7(+) nuclei remain unchanged 5 and 28 days after tmx administration, suggesting that TACC3 does not play a role in maintaining the quiescent satellite cell pool (Fig. 6A,B).
We therefore asked whether TACC3 is required for expansion of activated satellite cells in vivo. We used ctx to injure TA muscle of tmx-treated Pax7CreERT2/+; Tacc3fl/fl mice (Fig. 6C). Seven days after acute injury, we confirmed decreased numbers of TACC3(+) and PAX7(+) satellite cells (Fig. 6C-E) and the decrease in PAX7 expression is confirmed by western blotting of lysates of isolated satellite cells (Fig. 6F). To examine defects in proliferation in Pax7CreERT2/+; Tacc3fl/fl mice in vivo, we administered EdU to mice after ctx injury to TA muscle. Seven days after injury, we analyzed cross sections of regenerating TA muscle with EdU labeling combined with immunofluorescence with antibodies against PAX7 and myogenin. TA muscle of Pax7CreERT2/+; Tacc3fl/fl mice had numerous EdU(+) proliferating cells and differentiating myogenin(+) myogenic progenitors, but a marked decrease in PAX7(+) nuclei (Fig. S4). To examine specifically the satellite cell compartment of regenerating muscle, we used MACS to isolate myogenic progenitors from regenerating muscle. Although we did not observe significant differences in EdU incorporation amongst the total population of isolated myogenic progenitors (Fig. 6G,H), the PAX7-expressing compartment of EdU(+) cells is specifically reduced in Pax7CreERT2/+; Tacc3fl/fl mice (Fig. 6G,I). Similar results were revealed by analysis of TA sections with antibodies against PAX7 and Ki67 (Fig. 6J,K).
The inability of Tacc3-deficient satellite cells to expand in vivo should lead to a reduction in the pool of myogenic progenitors required for efficient muscle regeneration. Although Tacc3-deficient satellite cells differentiate to express myogenin (Fig. 6L,M; Fig. S4), we observe defective regeneration 7 days after injury, illustrated by weak, disorganized expression of embryonic myosin heavy chain (embMHC), and by large nuclei-dense areas devoid of regenerating myofibers (Fig. 6L,M; Fig. S4). The delay in regeneration mediated by Tacc3-deficient satellite cells continues 21 days after ctx injury, revealed by the presence of disorganized, smaller myofibers (Fig. 6N,O).
P-eIF2α-dependent Tacc3 translation permits satellite cell expansion ex vivo
In this work, we set out to understand how sal003-mediated inhibition of eIF2α dephosphorylation enables the expansion of satellite cells and further identify important regulators of satellite cell behavior during the expansion phase. Genetic manipulations that eliminate eIF2α phosphorylation cause satellite cells to activate, enter the cell cycle and contribute to differentiation, while being defective for self-renewal. The presence of sal003 in culture conditions leads to decreased rates of proliferation (Zismanov et al., 2016), and we propose that expansion occurs because slowly proliferating satellite cells avoid differentiating into post-mitotic myoblasts that undergo fusion to form the myofiber. Our data supports this model because genes required for differentiated skeletal muscle tissue are the most significantly downregulated in the presence of 10 µM sal003 (Figs 1,2).
Here, we asked whether sal003 enables satellite cell expansion by P-eIF2α-dependent translation of specific mRNAs. Culture of satellite cells in the presence of sal003 broadly leads to maintenance of gene expression associated with stem and progenitor cells (Figs 1,2). Amongst our candidates for P-eIF2α-dependent translation, enriched genes were most commonly associated with the spindle pole, centrosome and microtubule assembly (Fig. 3; Table S1), suggesting that sal003 permits the translation of mRNAs required to maintain the fidelity of cell division. Amongst these genes, we focused on Tacc3 because its 5′ UTR includes five uORFs and we demonstrate read-through of these inhibitory uORFs in a P-eIF2α-dependent manner. Moreover, Tacc3 depletion leads to a loss of PAX7 expression in satellite cells and precocious differentiation, whereas Tacc3 overexpression leads to increased expression of PAX7 in satellite cells. It remains unclear whether the changes in PAX7 expression that we observe in relation to perturbed TACC3 levels is indirectly associated with altering the balance of self-renewing versus differentiating satellite cells or mediated via direct mechanisms, for example interactions between PAX7 and TACC3. TACC3 has been reported to regulate the activity of transcription factors, for example TACC3 regulates subcellular localization and physical association of GATA1 and FOG1 (ZFPM1) to control hematopoiesis (Garriga-Canut and Orkin, 2004).
Satellite cell expansion and self-renewal ex vivo and in vivo requires TACC3
The efficiency of skeletal muscle regeneration is dependent on a balance between satellite cells undergoing self-renewal, proliferation and differentiation. Our results point to a role for TACC3 in permitting the expansion of PAX7(+) myogenic progenitors ex vivo and in vivo. Interestingly, the progeny of activated Tacc3-deficient satellite cells remain capable of limited proliferation, can activate the myogenic program and differentiate. However, the proliferating pool of PAX7-expressing myogenic progenitors is specifically depleted ex vivo and in vivo. The resulting balance between self-renewal and differentiation is perturbed, leading to defective muscle regeneration at early (7 days) and late (21 days) stages after acute injury.
Recently the regulation of stem cell polarity has emerged as an important mechanism facilitating stem cell expansion and differentiation. In satellite cells, an EGFR-AuroraA kinase signaling pathway orients the mitotic spindle apicobasally to facilitate asymmetric cell divisions that are required to expand myogenic progeny needed for efficient skeletal muscle regeneration (Wang et al., 2019). TACC3 has been identified as an EGFR binding partner, promoting EGFR stability at the cell surface to increase EGFR-dependent signaling pathways (Pettschnigg et al., 2017). Moreover, TACC3 is amongst several substrates of AuroraA kinase that influence mitotic spindle assembly (Burgess et al., 2015). Whether TACC3 stabilizes EGFR at the membrane of satellite cells and whether TACC3 regulates the orientation of the mitotic spindle required for satellite cell polarity are important questions that require further investigation.
In addition to the balance between satellite cell self-renewal, proliferation and differentiation, rates of apoptosis would impact satellite cell expansion ex vivo and in vivo. Germline deletion of Tacc3 has been reported to result in high levels of apoptosis in hematopoietic progenitors present in the thymus by embryonic day 18.5 (Piekorz et al., 2002). In our study, we did not detect increased rates of apoptosis in satellite cells in vivo 7 days after ctx injury, nor ex vivo after siRNA knockdown of Tacc3 (data not shown). These discrepancies may reveal a cell- or temporal-dependent context role for Tacc3 in apoptosis. Whether TACC3 protects against apoptosis in conditions of increased or extended proliferative stress, for example during embryonic development of muscle, chronic muscle degeneration or in aging satellite cells that are more prone to apoptosis, requires further investigation.
Increasing evidence points to translational control of gene expression as an important regulator of adult stem cell quiescence and self-renewal, with multiple mechanisms mediated by microRNA and RBPs repressing the translation of specific mRNAs. Here, we show increases in protein production, independent of mRNA levels, for a number of genes expressed in satellite cells cultured in the presence of sal003. We demonstrate P-eIF2α-dependent translation of mRNA for Tacc3 and show a requirement for Tacc3 expression in expanding satellite cells ex vivo and in vivo. Our findings suggest an additional role for P-eIF2α-dependent translation of mRNA in the maintenance of adult stem cell populations and outside the context of the ISR.
MATERIALS AND METHODS
Animal care practices were in accordance with the federal Canadian Council on Animal Care, as practiced by McGill University. All mice were maintained on a C57/Bl6 background. Tacc3fl/fl mice were kindly provided by R. Yao (Yao et al., 2016). Pax7CreERT2/+ (Murphy et al., 2011) and Rosa26tdTomato (Madisen et al., 2009) mice were obtained from Jackson Laboratories. Tmx (Cayman Chemical) was administered in corn oil, 30% ethanol by intraperitoneal injections (2.5 mg/day) for 5 days and, when indicated, mice were maintained on a tmx diet (80 mg/kg body weight/day, Envigo). For muscle regeneration, 8-week-old mice were anesthetized by isofluorane (CDMV) inhalation and 50 µl of 10 µM ctx (Sigma-Aldrich) was injected into the TA muscle. For 5-ethynyl-2′-deoxyuridine (EdU) labeling (Life Technologies), mice received 200 µg EdU in 100 µl PBS by intraperitoneal injections five times at 12-h intervals before analysis at day 7 after ctx injury. At indicated time points, muscle was harvested for analysis by immunofluorescence.
Cell and single-fiber isolation and culture
Satellite cells were isolated from abdominal and diaphragm muscle of 8-week-old Pax3GFP/+, Pax7CreERT2/+, R26tdtomato/+; Pax7CreERT2/+, R26tdtomato; Tacc3fl/fl mice by flow cytometry (GFP or tdTomato) as previously described (Zismanov et al., 2016) using a FACSAriaII cell sorter (BD Biosciences), or alternatively from Pax7CreERT2/+ or Pax7CreERT2/+, Tacc3fl/fl mice by MACS Satellite Cell Isolation Kit, together with anti-Integrin α-7 MicroBeads (Miltenyi). Single myofibers were isolated by trituration of 0.5% collagenase D (Sigma-Aldrich)-treated EDL muscle of 8-week-old adult mice. Cells and single EDL myofibers were cultured in 39% DMEM, 39% F12, 20% fetal calf serum (FCS) (Life Technologies), 2% UltroserG (Pall Life Sciences). When indicated, culture conditions were supplemented with 0.1% dimethylsulfoxide (DMSO control, Sigma-Aldrich), 10 µM sal003 (Sigma-Aldrich), 200 nM ISRIB (Cayman), 10 µM 4-OHT (Cayman) or 2.5 µM KHS101 (Sigma-Aldrich). For siRNA experiments, satellite cells were transfected after 48 h culture with siRNAs against Tacc3 (20 nM) or Mission® siRNA Universal Negative control (Sigma-Aldrich) with jetPRIME® transfection reagent (Illkirch-Graffenstaden, France). Transfected satellite cells were cultured for an additional 2 days. For EdU incorporation assays (Life Technologies), 10 µM EdU was included in culture for 2 h or 6 h, when indicated. We cultured 293 cells in 90% DMEM, 10% FCS and supplemented when indicated with 1 µM TG (Sigma-Aldrich).
The mouse Tacc3 sequence was amplified by PrimeSTAR® Max DNA polymerase (TAKARA) using forward 5′-CTCCCCAGGGGGATCATGAGTCTGCATGTCTTAAAT-3′ and reverse 5′-GAGGTTGATTGTCGATCAGATCTTCTCCATCTTAG-3′ primers. The purified fragment was cloned into BamHI-SalI site of pLenti-PGK-GFP (Crist et al., 2009) using In-Fusion cloning HD kit (TAKARA). HEK293T cells were transfected with the pLenti-PGK-Tacc3 or pLKO.1 TRC (control, gift from David Root, Broad Institute of MIT and Harvard, USA, Addgene plasmid #10879), pMD2.G (gift from Didier Trono, École polytechnique fédérale de Lausanne, France, Addgene plasmid #12259) and psPAX2 (gift from Didier Trono, Addgene plasmid #12260) with jetPRIME® transfection reagent (Illkirch-Graffenstaden). Six hours later, the medium was changed and 42 h later, supernatant was collected, filtrated through a 0.45 µm filter and concentrated using a Lenti-XTM Concentrator (TAKARA). Titers (∼1×108 infectious units/ml) were calculated by GFP analysis of transduced 293T cells. Satellite cells isolated by either MACS or FACS described above (1×104 cells/35 mm dish or a well of a 6-well plate) were transduced with 4 µl of lentivirus solution with polybrene (5 µg/ml). Twenty-four hours after transduction, the lentivirus-containing medium was carefully removed and replaced with fresh satellite cell medium. Transduced satellite cells were cultured for an additional 3 days.
The Atf4- and Tacc3-luciferase constructs were made by cloning gBlock gene fragments (Integrated DNA Technologies) corresponding to 5′ UTRs of Mus musculus Atf4 and Tacc3, fused to the 5′ end of the firefly luciferase (fluc) gene up to the NarI restriction enzyme site. Mutant versions of Atf4 and Tacc3 5′ UTRs were designed with each ATG start codon of uORFs deleted. These 5′ UTR gene fragments were cloned into HindIII, NarI sites upstream of the fluc gene in pGL3-promoter plasmids, thereby maintaining the final overlapping uORF with the main ORF for fluc (Promega). HEK293 cells were plated in 24-well plates at a density of 25,000 cells/well and incubated overnight. Fluc reporter plasmids and the pRL-TK Renilla luciferase (Rluc) transfection control plasmid (Promega) were co-transfected into these cells using jetPRIME® transfection reagent (Illkirch-Graffenstaden). Cells were incubated overnight in the presence or absence of 1 µM TG (Sigma-Aldrich). Cells were then lysed and the Fluc/Rluc ratio was determined using the Promega Dual-Luciferase Reporter kit.
Cultured satellite cells were fixed in 4% paraformaldehyde (PFA), permeabilized with 0.2% Triton, 50 mM NH4Cl and blocked in 5% horse serum (HS). Single EDL myofibers were fixed with 4% PFA, permeabilized with 0.1% Triton in PBS and blocked in 5% HS with 0.1% Triton in PBS. TA muscles were fixed for 2 h in 0.5% PFA at 4°C and equilibrated overnight in 20% sucrose at 4°C. Tissues were mounted in Frozen Section Compound (VWR) and flash frozen in a liquid nitrogen-cooled isopentane bath. Transverse sections (10 µm) were permeabilized with 0.1% Triton, 0.1 M Glycine in PBS, and blocked in M.O.M. reagent (Vector Laboratories). For immunoblotting, cell lysates were obtained in RIPA buffer (Thermo Fisher Scientific) supplemented with Complete protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Sigma-Aldrich).
Primary antibodies were against PAX7 [Developmental Studies Hybridoma Bank (DSHB) 1:100], MYOD (Santa Cruz Biotechnology, sc-304 and sc-377460, 1:300), TACC3 (Abcam, 134154, 1:200), Ki67 (BD Biosciences, B56, 1:300), laminin (Sigma-Aldrich, L9393, 1:500), myogenin (Abcam, 124800, 1:200), myogenin (Santa Cruz Biotechnology, sc-12732, 1:200), embMHC (DSHB, F1.652, 1:100) and β-actin (Sigma-Aldrich, A5441, 1:2000). Alexa Fluor-488, Alexa Fluor-594 and Alexa Fluor-647 conjugated secondary anti-mouse IgG1, anti-mouse IgG and IgG2b or anti-rabbit antibodies (Life Technologies, A21121, A21207, A21145 and A31573, 1:500) were used for immunofluorescence. Horseradish peroxidase (HRP)-conjugated anti-mouse or anti-rabbit secondary antibodies (Jackson ImmunoResearch, 115-035-003 and 111-035-003, 1:2000) were used with the ECL Prime Western Blotting Detection reagents (GE Healthcare). Densitometry of immunoblots and total cell fluorescence analyses were performed using ImageJ.
RNA was isolated from cells with TRIzol reagent (Life Technologies) and treated with DNase (Roche) before reverse transcription with iScript reverse transcription supermix (Bio-Rad). RT-PCR primers were: Pax7 forward 5′-AGGCCTTCGAGAGGACCCAC-3′, reverse 5′-CTGAACCAGACCTGGACGCG-3′; myogenin forward 5′-CAACCAGGAGGAGCGCGATCTCCG-3′, reverse 5′-AGGCGCTGTGGGAGTTGCATTCACT-3′; Tacc3 forward 5′-GAGCTTCAGAGACCCATCAGA-3′, reverse 5′-AGTTGGAGAGATGGGACGAG-3′; Actb forward 5′-AAACATCCCCCAAAGTTCTAC-3′, reverse 5′-GAGGGACTTCCTGTAACCACT-3′. Levels of mRNA were measured using SYBR Green on a 7500 Fast Real Time PCR System (Applied Biosystems).
RNA-seq, mass spectrometry and bioinformatic analysis
For both RNA and protein analysis, satellite cells were isolated from Pax3GFP/+ adult mice and seeded at 7500 cells per 35 mm plate and subsequently cultured in 10 µM sal003 (n=4) or DMSO (n=4) for 4 days. For RNA analysis, total RNA was isolated from satellite cell cultures using the RNeasy® Micro Kit (Qiagen). The RiboGone™-mammalian (Takara Bio) kit was subsequently used to eliminate ribosomal and mitochondrial RNA. cDNA libraries for RNA-seq were then prepared using the SMARTer® Stranded RNA-Seq kit (Takara Bio). With all kits, the manufacturer's protocols were followed. The resulting cDNA libraries were sequenced with the Illumina MiSeq system (Genome Quebec).
For protein analysis, the TMTsixplex™ Isobaric Tag Kit (Thermo Fisher Scientific) was used to isolate and label peptides from satellite cells cultured in 10 µM sal003 (n=12 plates) or DMSO (n=12 plates), per the manufacturer's protocol. After labeling peptides with unique isobaric tags, the peptide solutions were pooled and analyzed via high pH reversed phase fractionation and liquid chromatography-mass spectrometry. All data were acquired with Thermo Orbitrap Fusion™ Tribrid™ 2.1 software (Thermo Fisher Scientific), analyzed using Proteome Discoverer 1.4 (Thermo Fisher Scientific) and MASCOT v2.4 software (Matrix Science). Raw data files were searched against a Uniprot Mouse database. Gene lists were examined using Enrichr (Chen et al., 2013) for ENCODE and ChEA consensus transcription factor binding sites, ENCODE histone modifications, mouse tissue expression, GO Biological Process and GO Cellular Component. Venny2.1.0 was used to generate a list of genes upregulated at the protein level (log2>1), without a corresponding increase in mRNA expression (log2<1).
Trimmomatic v0.32 (Bolger et al., 2014) was used to trim sequencing reads, including adaptors and other Illumina-specific sequences, the first four bases from the start of each read and low quality bases identified using a 4 bp sliding window where quality fell below 30 (phred33<30). Finally, reads shorter than 30 base pairs were removed. Cleaned reads were aligned to the mouse reference genome build mm10 using STAR v2.3.0e (Dobin et al., 2013) with default settings. Reads mapping to more than 10 locations in the genome (MAPQ<1) were discarded. Gene expression levels were estimated by quantifying uniquely mapped reads to exonic regions (the maximal genomic locus of each gene and its known isoforms) using featureCounts (v1.4.4) (Liao et al., 2014) and the Ensembl gene annotation set. Normalization (mean of ratios) and variance-stabilized transformation of the data were performed using DESeq2 (v1.14.1) (Love et al., 2014). Multiple control metrics were obtained using FASTQC (v0.11.2), samtools (v0.1.20) (Li et al., 2009) BEDtools (v2.17.0) (Quinlan and Hall, 2010) and custom scripts. For visualization, normalized Bigwig tracks were generated using BEDtools and UCSC tools. PCA was carried out using the 1000 most variant genes.
Graphs are presented as mean±s.e.m., as indicated in figure legends. Unless otherwise indicated, three independent replicates of each experiment were performed. Significance was calculated by unpaired Student's t-tests with two-tailed P values: *P<0.05, **P<0.01, ***P<0.001.
We thank C. Young for assistance with flow cytometry, C. Borchers and D. Smith for assistance with mass spectrometry. R. Yao generously provided Tacc3fl/fl mice. Data analyses were enabled by computer and storage resources provided by Compute Canada and Calcul Québec.
Conceptualization: C.C.; Methodology: R.F., S.J., G.L., C.L.K., C.C.; Software: S.H., C.L.K.; Validation: C.C.; Formal analysis: R.F., S.J., G.L., S.H., C.L.K.; Investigation: R.F., S.J., G.L., H.C.M.C., S.H., C.L.K.; Data curation: S.H., C.L.K.; Writing - original draft: C.C.; Writing - review & editing: R.F., S.J., G.L., H.C.M.C., C.C.; Visualization: R.F., S.J., G.L., H.C.M.C., C.C.; Supervision: C.L.K., C.C.; Project administration: C.C.; Funding acquisition: C.C.
C.C. and coworkers are funded by the Canadian Institutes of Health Research (CIHR; 399258), the Stem Cell Network, the Fonds de Recherche du Québec – Santé (FRQS) and the Richard and Edith Strauss Foundation. R.F. is funded by a Japan Society for the Promotion of Science Overseas Research Fellowship, the Uehara Memorial Foundation, the Mochida Memorial Foundation for Medical and Pharmaceutical Research, and a Ministry of Education, Culture, Sports, Science and Technology Leading Initiative for Excellent Young Researchers grant. C.L.K. is supported by the CIHR (156086) and the FRQS.
RNA-sequencing data have been deposited in GEO under accession number GSE164774.
Peer review history
The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.194480.reviewer-comments.pdf
The authors declare no competing or financial interests.