Abstract
Phenylalanyl tRNA synthetase (PheRS) levels are elevated in multiple cancers and, interestingly, also in normal stem cells. Our results show that elevated levels of the α-PheRS subunit stimulate cell proliferation in different tissues, while downregulation of α-PheRS reduces organ size. Furthermore, overexpression of α-PheRS in stem- and progenitor cells caused over-proliferation in the intestine, a phenotype indistinguishable from the Notch RNAi phenotype in the same cells. Importantly, the phenotype caused by high levels of α-PheRS can be rescued by simultaneously overexpressing Notch, suggesting that α-PheRS induces this phenotype by downregulating Notch. High levels of α-PheRS in neuroblasts also cause the same phenotype as knocking down Notch in these cells, even though Notch signaling in the neuroblast lineage serves an opposite function by promoting neuroblast proliferation and maintenance. α-PheRS might, therefore, act as a general novel regulator of Notch signaling. α-PheRS levels, in turn, are controlled by Stat92E, the transcription factor of the JAK/STAT signaling pathway that is needed for the differentiation of intestinal stem cells during normal tissue homeostasis. From this, we conclude that the α-PheRS subunit can transmit the activity status of the JAK/STAT pathway to the Notch pathway as a mechanism to coordinate stem cell proliferation with differentiation. In this process, α-PheRS levels balance between tissue development and tissue growth to regulate tissue homeostasis. For its established essential function as an aminoacyl tRNA synthetase, α-PheRS needs to bind to β-PheRS in every cell to form the α2β2 heterotetramer that loads the amino acid phenylalanine onto the cognate tRNAPhe. Here we also demonstrate that the newly identified activities of α-PheRS are moonlighting functions, independent of the aminoacylation activity of PheRS, and they do not visibly stimulate translation.
Introduction
Many cancer tissues display higher levels of Phenylalanyl-tRNA synthetase (PheRS, FARS or FRS) than their healthy counterparts according to the database “Gene Expression across Normal and Tumor tissue” (GENT2; published 11 July 2019). Interestingly, a correlation between tumorigenic events and PheRS expression levels had been noted already much earlier for the development of myeloid leukemia (Sen et al., 1997). Despite this, a possible causative connection between elevated PheRS levels and tumor formation had so far not been reported and, to our knowledge, also not been studied. This might have been due to the assumption that higher PheRS levels could simply reflect the demand of tumorigenic cells for higher levels of translation, or it could have to do with the difficulty of studying the moonlighting function of a protein that is essential in every cell for basic cellular functions such as translation.
Aminoacyl-tRNA synthetases (aaRSs) are important enzymes that act by charging transfer RNAs (tRNAs) with their cognate amino acid, a key process for protein translation. This activity makes them essential for accurately translating the genetic information into a polypeptide chain (Schimmel and Soll, 1979). Besides their well-known role in translation, an increasing number of aaRSs have been found to perform additional functions in the cytoplasm, the nucleus and even outside of the cell (Guo and Schimmel, 2013; Nathanson and Deutscher, 2000; Smirnova et al., 2012) (Casas-Tinto et al., 2015; Gomard-Mennesson et al., 2007; Greenberg et al., 2008; Otani et al., 2002; Zhou et al., 2014). Moonlighting aaRSs regulate alternative splicing, RNA processing and angiogenesis (Lee et al., 2004). For example, the amino-acid binding site of LysRS has an immune response activity; or TrpRS inhibits the vascular endothelial (VE)-cadherin, which elicits an anti-angiogenesis activity (Tzima et al., 2005; Yannay-Cohen et al., 2009).
Cytoplasmic PheRS is one of the most complex members of the aaRSs family, a heterotetrameric protein consisting of 2 alpha-(α) and 2 beta (β)-subunits responsible for charging tRNAPhe during translation (Roy and Ibba, 2006). The α subunit includes the catalytic core of the tRNA synthetase and the β subunit has structural modules with a wide range of functions, including tRNA anticodon binding, hydrolyzing mis-activated amino acids, and editing misaminoacylated tRNAPhe species (Ling et al., 2007; Lu et al., 2014; Roy and Ibba, 2006). Precision of the initial charging reaction and proper editing are both important for the cell and the organism because mutations in the amino-acid recognition and editing sites of Drosophila PheRS cause sieving defects that lead to mis-incorporation of amino acids into proteins, ER stress, apoptosis, shortened life span, as well as neural degeneration (Lu et al., 2014).
We set out to address the question whether and how elevated levels of PheRS can contribute to tumor formation. To test for this activity, we studied the role of PheRS levels in the Drosophila model system with the goal of dissecting the molecular mechanism of such a moonlighting role of PheRS. We found that α-PheRS levels regulate cell proliferation, cell differentiation or both in different tissues and cell types. We now show that elevated levels of a-PheRS do not simply act to allow higher levels of translation, but control signaling mechanisms involved in differentiation and proliferation control. Although the consequences of altered levels vary to some degree between tissues, we found that even in two tissues with the most divergent consequences, PheRS levels acted by repressing the Notch signaling pathway, suggesting that this regulative mechanism is responsible for all moonlighting activities of α-PheRS described here. α-PheRS levels, in turn, are controlled by JAK/STAT signaling in the intestinal system, placing α-PheRS at the intersection between two signaling pathways for the fine tuning of normal tissue homeostasis in the midgut. Focusing on the intestine and intestinal stem cells revealed that elevated α-PheRS levels are tumorigenic in the intestinal model. Given the high demand for research on intestinal diseases and cancer (Markowitz and Bertagnolli, 2009) our work now opens new avenues to test ways to control tissue homeostasis and tumor formation.
Results
A non-canonical α-PheRS activity is sufficient to induce additional M-phase cells
To test whether elevated levels of PheRS can stimulate growth or proliferation when other aaRSs are not overexpressed, we overexpressed α-PheRS, β-PheRS, and both subunits together in the posterior compartment (P) of wing discs using the engrailed-Gal4 (en-Gal4) driver. In this assay the anterior compartment expresses normal endogenous levels and serves as an internal control. When α-PheRS was overexpressed in the posterior compartment either alone or together with β-PheRS, the mitotic marker phospho-histone H3 (PH3) revealed a 40% increase in mitotic cells in the posterior (P) compartment relative to the anterior (A) one of the same disc (Fig 1A-C). Because elevated levels of the α-PheRS subunit alone are sufficient for the increase in mitotic cells, this effect is unlikely caused by increased tRNAPhe aminoacylation activity and translational activity, which requires both subunits.
To test this interpretation, we made a mutant version of α-PheRS in which Tyr412 and Phe438 are replaced by Cysteins. These substitutions are predicted to block the entrance into the phenylalanine binding pocket, preventing binding of Phe and aminoacylation of tRNAPhe by the mutant PheRSCys (Finarov et al., 2010). To test whether the PheRSCys substitution indeed reduces the aminoacylation activity of PheRS, we expressed mutant and wild-type α-PheRS subunits together with β-PheRS subunits in E. coli, purified them and performed aminoacylation assays. As opposed to the strong enzymatic activity of the wild-type α-PheRS plus β-PheRS subunits, the α-PheRSCys together with wild-type β-PheRS produced only the same background signal as the α-PheRS subunit alone (Supplementary Fig S1). The same mutations were also introduced into a genomic clone and the resulting transgenic α-PheRSCys was not able to rescue the α-PheRSG2060 mutant, indicating that the Cys mutant is indeed not functional in aminoacylation in vivo in Drosophila. Overexpressing a transgenic α-PheRSCys in the posterior compartment of the wing disc also caused a 67% increase in the number of mitotic cells in the above assay (Fig 1C). The fact the mutant α-PheRSCys version caused an increase in mitotic cells at least as strongly as the wild-type α-PheRS, together with the fact that β-PheRS overexpression was not needed for this effect, clearly demonstrates that the canonical function of PheRS is not required to cause the elevated frequency of mitotic cells.
We also tested directly whether PheRS overexpression is unable to cause elevated translation as we expected. For this, we analyzed general protein synthesis activity in the two wing compartments by puromycin staining using the ribopuromycylation method (RPM) (Deliu et al., 2017). Overexpression of the transcription factor dMyc increases protein synthesis activity and was therefore used as a positive control for detection of elevated translation and PMY labeling (Deliu et al., 2017). Indeed, when comparing signal intensity in the posterior compartment to the intensity in the anterior compartment of the same disc, dMyc overexpression significantly increased the anti-PMY signal in the expressing posterior compartment. In contrast, neither overexpression of α-PheRS alone nor combined with β-PheRS increased the puromycin labeling in the overexpressing compartment (Fig 1D-E’’, F). The combined results therefore demonstrate unambiguously that elevated α-PheRS levels cause additional cells to be in mitosis through an aminoacylation- and translation-independent, non-canonical activity. α-PheRS levels might specifically slow down progression through M-phase, causing higher numbers of cells to remain in the PH3-positive state. Alternatively, they might either promote over-proliferation of mitotic cells or induce proliferation in non-cycling cells.
High α-PheRS tolerance in an organ with tight size control
To find out whether the increased number of mitotic cells in the region expressing higher α-PheRS levels leads to a different compartment size, we measured the size of two clearly defined regions in the posterior (P) and the anterior (A) compartment of the adult wing as shown in Fig 2A. Normalizing the size of the posterior region (P) to the size of the anterior region (A), we calculated the P/A ratio (Fig 2B). Even though α-PheRS alone was sufficient to induce additional mitotic cells in larval discs, this did not cause the formation of a larger wing in the adult. Co-overexpression of both subunits was needed to cause a small, but significant increase of the posterior region of the wing (Fig 2B). This size increase does not appear to be a general property of elevated aaRS levels because overexpression of Glycyl-tRNA synthetase (GlyRS; (Niehues et al., 2015)) with the same driver did not increase wing size (Fig 2B). The final size of this particular organ is tightly controlled by different mechanisms that are only partially understood, but are capable of compensating for differences in growth and proliferation such that the compartments reach their correct final size even if they grew at different rates at an earlier stage (Hariharan, 2015; Martin and Morata, 2006). To better understand the consequences of high PheRS levels in the discs, we studied its effect on dissociated cells and found that elevated PheRS levels primarily affected cell size, whereas proliferation remained controlled or was compensated for (Supplementary data, Fig S2A,B).
We considered the possibility that PheRS might signal availability of Phe to TORC1, which links cell growth to amino acid availability (Laplante and Sabatini, 2012; Wullschleger et al., 2006). Such a mechanism would be analogous to the function of LeuRS in this pathway (Bonfils et al., 2012; Han et al., 2012). However, experimental testing of this hypothesis did not uncover any evidence for such a signaling function (Supplementary data, Fig S3A,B).
The α-PheRS subunit accelerates proliferation in different tissues and its knockdown reduces organ size
To test the effects of PheRS levels on proliferation directly and in a different organ and cell type, we set up a mosaic analysis with repressible cell marker (MARCM; (Wu and Luo, 2006) assay in the follicle cells. Twin spot clones were generated with one clone overexpressing PheRS and the GFP marker, and its twin clone expressing normal endogenous levels of PheRS and serving as an internal control (Fig 3A). The results of this experiment showed that clonal overexpression of both subunits of PheRS accelerated the proliferation of the overexpressing cells on average by 32% (Fig 3B). Overexpression of GFP with only the β-PheRS subunit or with GlyRS did not significantly promote clonal expansion (Fig 3B), confirming that the activity of stimulating growth and proliferation is specific for PheRS and not a general role of aaRSs. Interestingly, overexpression of GFP with the α-PheRS subunit alone also stimulated cell proliferation autonomously by 30% (Fig 3B) and, intriguingly, this was very close to the 32% increase calculated for the clone overexpressing both PheRS subunits (Fig 3B). Remarkably, the higher number of mitotic cells observed upon α-PheRS overexpression in the posterior compartment of the larval wing discs (Fig 1C) was in a comparable range as the proliferation increase in the follicle cell assay (Fig 3B). These results therefore show that α-PheRS levels promote cell proliferation and they suggest that it has this activity in different tissues.
Overexpression of α-PheRSCys alone and α-PheRSCys together with β-PheRS (PheRSCys) stimulated clonal growth and cell proliferation in the follicle cell twin spot experiment by 28% and 25%, respectively (Fig 3B), again confirming that this is an aminoacylation-independent activity. It is tempting to speculate that the slightly lower increase in proliferation upon overexpressing the mutant α-PheRSCys (28%) in twin spot clones compared to wild-type α-PheRS (30%) could be due to reduced translation caused by the overexpression of an inactive α subunit that is predicted to partially act as a dominant negative subunit for the aminoacylation function. If this proliferation function is important for development and homeostasis, reduced α-PheRS activity should lead to problems in these processes. Because α-PheRS and β-PheRS are both essential genes in Drosophila (Lu et al., 2014), we used RNAi to reduce their activity in specific tissues by RNAi (Fig S4A,B). Indeed, knock down in the developing eye reduced the adult eye, whereas knockdown in the larval fat body reduced the size of the entire pupae. Furthermore, RNAi knockdown α-PheRS or β-PheRS in Kc cells caused these cells to proliferate more slowly than the control cells (Fig S4C). Although it is not clear from these results, which activity of PheRS causes this effect, the phenotypes observed are consistent with a direct function in regulating proliferation.
α-PheRS promotes stem cell proliferation in the midgut and high expression leads to hyper- and dysplasia
Tissue growth and homeostasis play important roles in developing and outgrown animals, and they require tight control of stem cell self-renewal and differentiation of daughter cells. The Drosophila midgut is a powerful model to analyze these mechanisms and their interplay. Intestinal stem cells (ISCs), also referred to as adult midgut progenitors (AMPs) in the larval gut, can either divide asymmetrically or symmetrically. After an asymmetric division, one daughter cell differentiates into an absorptive enterocyte (EC) or a secretory enteroendocrine (EE) cell, the other keeps its stem cell identity (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). We investigated the effect of overexpression of each PheRS subunit in larval ISCs by using the esg-Gal4 driver, which expresses specifically in ISCs. Co-expression of UAS-YFP allowed us to monitor the esg-Gal4 activity. The results demonstrated that overexpression of Myc::α-PheRS alone in the larval midgut caused the numbers of YFP positive ISC cell clusters in the posterior midgut to increase. In addition, the number of YFP positive cells per cluster increased as well (Fig 4A, B, D-D’). The increase of both numbers was significant when compared to posterior midguts expressing normal levels of α-PheRS (Fig 4A, B, C-C’). Overexpression of β-PheRS::V5 alone did not produce such a phenotype (Fig 4A, B, E-E’). Interestingly, when both PheRS subunits were overexpressed, the guts also showed an increase in the number of YFP positive ISC clusters. However, the increase was less pronounced and the number of YFP positive cells per cluster remained unchanged (Fig 4A, B, F-F’).
Driving the expression of the α-PheRSCys mutant alone in ISCs also produced additional ISCs, indicating again that the aminoacylation function is not required for this activity (Fig 4G). Surprisingly, however, this treatment induced the over-proliferative phenotype in both anterior and posterior areas of the larval midgut (outlined with white dashed lines, Fig 4G, G’) while the overexpression of wild-type Myc::α-PheRS gave rise to high numbers of ISCs only in the posterior midgut (Fig 4D, D’). Interestingly, elevated α-PheRSCys levels also caused the appearance of a more severe, tumor-like phenotype, where individual ISC clusters could not be discerned anymore. Furthermore, instead of the wild-type gut phenotype, characterized by a majority of ECs with large nuclei, interspersed with occasional ISC clusters with smaller nuclei (as seen in the YFP overexpression control, Fig 4C, C’), we observed a phenotype where ECs and ISCs could not be distinguished based on the size of their nuclei, but emerged as a larger cell population with intermediate size nuclei (Fig 4G’). Many of these cells expressed the esg>YFP stem cell marker at high levels, but others displayed only a very weak YFP signal. One possible interpretation of this phenotype could be that α-PheRSCys overexpressing ISCs progress through the cell cycle more rapidly such that they are not able to grow to their proper size and do not have sufficient time to turn over the YFP. Staining these guts for the mitotic marker PH3 demonstrated that the posterior midgut contained clearly more and a higher proportion of PH3 positive cells. Significantly, more cells were labelled by anti PH3 staining in the posterior midgut when α-PheRS or the α-PheRSCys mutant were overexpressed alone in ISCs (Fig 4H, I). Again, this suggests that cells in these areas display an elevated proliferation rate.
In a normal adult midgut, ISCs are found as characteristic single cells or as pairs with their daughter enteroblast (EB). Both cell types are labeled with YFP when their expression is driven by esg-Gal4 (Fig 4J). Overexpression of only α-PheRS or α-PheRSCys in these adult ISCs caused a strong phenotype, too. It induced hyperplasia and dysplasia in regions R4-R5 of the posterior midgut. Similar to the hyperplasia phenotype observed in the larval gut, we observed cells adjacent to ISC clusters displaying the YFP stem cell marker, even though they contained large nuclei (Fig 4J’-J”), indicative of a dysplasia phenotype. We conclude that in the larval and adult guts, elevated α-PheRS expression can elevate the proliferation rate of stem cells and lead to hyper- and dysplasia.
Elevated α-PheRS levels prevent proper differentiation and gut homeostasis by downregulating Notch signaling
Asymmetric divisions of ISCs give rise to a new ISC and an undifferentiated EB. Differential Delta/Notch signaling between the new ISC and the EB causes the latter to either differentiate into an absorptive EC or a secretory EE (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). To investigate how α-PheRS affects the fate decision in this lineage, we studied the cell population in the posterior midgut. In larval and adult guts, the over-proliferation phenotype caused by α-PheRSCys overexpression showed a significantly elevated ratio between EEs and ECs (Fig 5A-A’, B-B’, C). The overexpression of α-PheRS alone with the esg-Gal4 driver caused the same changes, albeit with a slightly lower expressivity (Fig 5C). These observations indicate that overexpression of α-PheRS or α-PheRSCys directs EB differentiation to the EE fate and interferes with EC differentiation.
Differentiating ISC daughter cells, EBs, adopt the EE fate when their Notch activity is low and the EC fate if their Notch activity is high (Takashima et al., 2011). Furthermore, because Notch activity is needed for EB differentiation, reduced Notch activity leads to EE-like and ISC-like tumors (Micchelli and Perrimon, 2006; Patel et al., 2015; Wang et al., 2015; Yin and Xi, 2018). Indeed, in regions where the normal cellular composition of the larval gut was transformed, we observed an increase in the number of cells and in particular a cell population with intermediate size nuclei upon RNAi treatment of Notch with the esg-Gal4 driver (white dashed line, Fig 5D-D’). The similarity between the two phenotypes suggests that overexpression of α-PheRS or α-PheRSCys might cause this phenotype by reducing Notch activity. Indeed, as seen in Fig 5F, overexpression of α-PheRS and α-PheRSCys downregulates the Notch activity reporter NRE-EGFP (Notch response element promoter driving the expression of eGFP; (Housden et al., 2012) and overexpressing Notch together with α-PheRSCys in the same midgut ISCs rescued the tumor-like midgut phenotype, giving rise to midguts with a wild-type appearance (Fig 5A”, B”, E-E’). This shows that α-PheRS levels control Notch activity and that high α-PheRS levels cause the tumor-like phenotype in the midgut by downregulating Notch activity.
α-PheRS is a novel general repressor of Notch signaling
Developing larval brains contain neuroblast (NB) stem cells that divide asymmetrically to either keep the stemness and or to differentiate into neuronal cells. Notch signaling plays a crucial role in this process (Ables et al., 2011; Giachino and Taylor, 2014), but in contrast to the situation in the gut, loss of Notch prevents NB self-renewal, and ectopic expression of Notch leads to tumor formation (de la Pompa et al., 1997; Grandbarbe et al., 2003; Hatakeyama et al., 2004). This opposite role of Notch makes the NB lineage an ideal complementary system to test whether α-PheRS is a general component of the Notch pathway. Driving α-PheRS or the α-PheRSCys expression in NBs with the inscutable-Gal4 (incs-Gal4) driver resulted in significantly smaller central brains (CB), the region where the NBs are located (Fig 6A-A’’, B). In contrast, this treatment had little or no effect on the size of the optic lobes (OL). The phenotype was indistinguishable from the phenotype caused by Notch knock-down (Fig 6B), and co-overexpression of Notch with α-PheRSCys in neuroblasts rescued the brains to wild-type size (Fig 6B).
Type II neuroblasts are particularly suited to analyze effects on neuroblast differentiation. Targeting specifically the 8 type II neuroblasts in central brain lobes with α-PheRS or α-PheRSCys, overexpression, resulted in half of the central brains in a strong reduction of the number of neuroblasts, and the resulting central brain lobes ended up smaller (Fig 6C and Supplementary Fig S5). Knocking down Notch in type II NBs with RNAi and the same driver combination resulted in a similar phenotype, but with a higher expressivity. On the other hand, overexpression of Notch in α-PheRS overexpressing type II neuroblasts partially rescued the number of type II neuroblasts to wild-type numbers and it restored the normal size of the brain (Fig 6C and Supplementary Fig S5). Because the effect on Notch signaling is the same in guts and brains, two tissues where Notch has opposing functions on cellular differentiation, we conclude that α-PheRS is a novel general repressor of Notch signaling.
Transcription factor Stat92E of the JAK/STAT pathway regulates α-PheRS levels
Under normal conditions, the JAK/STAT signaling pathway induces progenitor differentiation into ECs via regulating Notch activity (Herrera and Bach, 2019). Under stress conditions like bacterial infection, secreted Unpaired3 (Upd3) acts as a cytokine that activates JAK/STAT signaling, leading to ISC proliferation and differentiation to repair the damaged parts of the gut (Buchon et al., 2009; Lin et al., 2010). Upd3 not only activates JAK/STAT signaling, but also Notch activity to enhance ISC proliferation and to promote EC differentiation, respectively (Jiang et al., 2009). Because α-PheRS levels affect Notch signaling, too, we therefore tested whether α-PheRS could possibly coordinate JAK/STAT with Notch signaling. The Stat92E gene encodes the transcription factor at the downstream end of the JAK/STAT pathway. Using the hypomorphic Stat92EHJ allele, we tested whether suppressing JAK/STAT signaling affects α-PheRS expression. Indeed, Stat92EHJ mutants displayed elevated levels of α-PheRS in Delta-positive ISC cells (Fig 7A-B”). Consistent with this result, Stat92E knock-down by RNAi with the esg-Gal4 driver also showed elevated α-PheRS levels in ISC cells (Fig7C-C”). As shown in Fig 7 D-D’, E-E’, F-F’, G, the levels of α-PheRS significantly increased in ISCs/EBs. The fact that two different approaches to reduce Stat92E activity lead to higher α-PheRS levels in ISCs/EBs clearly demonstrates that Stat92E regulates α-PheRS levels in these cells.
Surprisingly, RNAi knock down of Stat92E with the esg-Gal4 driver not only affected α-PheRS levels in ISCs, but also in some neighboring polyploid ECs, but not in distant ECs (arrows in Fig7C”, F) (Fig7G). Because we also observed this phenotype in Stat92EHJ mutants (arrows in Fig 7E) (Fig 7G), this is unlikely to be an off-target effect of the RNAi. It thus appears that Stat92E can normally control α-PheRS levels also in a cell non-autonomous way. To test if Stat92E is not only required for downregulation of α-PheRS, but also sufficient, we overexpressed Stat92E in ISCs and EBs. Indeed, elevated Stat92E levels reduced the α-PheRS signal in ISCs and EBs (Fig 7H-I”). On the other hand, high levels of α-PheRS did not affect Stat92E activity when assayed with the 10X STAT92E-GFP reporter (Bach et al., 2007) (Fig 7J-K”’). We conclude that Stat92E can regulate α-PheRS levels specifically in EBs and ISCs to maintain stem cell homeostasis. Together with the result that α-PheRS regulates Notch signaling, we therefore identified α-PheRS as an intermediate factor that links the JAK/STAT pathway to Notch signaling to regulate gut homeostasis.
Discussion
Our work revealed that PheRS not only charges tRNAs with their cognate amino acid Phe, but that it also performs moonlighting functions in regulating cell proliferation and differentiation in different tissues. Levels of α-PheRS are critical for these regulative processes and these levels are generally elevated in healthy stem cells compared to differentiated cells. Similarly, many tumor cells show elevated α-PheRS levels compared to their healthy counterparts and a positive correlation between these levels and tumorigenic events had been noted quite some time ago (Sen et al., 1997). Several circumstantial and direct evidence show that elevated α-PheRS levels do not simply allow higher translational activity to overcome a growth rate restriction imposed by hypothetically limiting levels of PheRS. In fact, PheRS is unlikely to be rate-limiting for cellular growth. because animals with only one copy of the α-PheRS or β-PheRS gene do not show a phenotype (Lu et al., 2014) and tissue culture cells can be stimulated to grow more rapidly without stimulating the expression of the two PheRS genes (Chen et al., 2003) (Lu et al., 2014). Indeed, directly measuring translational activity in situ (Fig 1D-D”) showed the same levels of translation whether α-PheRS was overexpressed or not. Furthermore, aminoacylation of tRNAPhe requires the tetrameric protein α2β2-PheRS. However, overexpression of an aminoacylation-dead α-PheRSCys mutant subunit alone, (without simultaneous overexpression of the β-PheRS subunit) already lead to the accumulation of numerous additional dividing cells, closely resembling the phenotype observed when the wild-type gene was expressed in the same way. We therefore conclude that this activity of α-PheRS is independent of the translational function of PheRS.
The notion that the α-PheRS subunit can be stable and it functions independently of the β-subunit was surprising because previous results showed that the two subunits were dependent on the presence of the other subunit for their stability (Antonellis et al., 2018; Lu et al., 2014).
Our results now show that this requirement does not apply to all cell types. In young follicle cells, ISCs and possibly other dividing cells, the overexpression of the α-PheRS subunit alone results in higher levels of α-PheRS accumulation and, in particular when elevated levels were induced in gut ISCs and EBs, this produced a strong phenotype. This suggests that the α- and β-PheRS subunits function together in every cell to aminoacylate tRNAPhe, but in addition, the α-subunit can be stable in specific cell types, such as stem cells, where it assumed a novel function in regulating cell proliferation and differentiation.
PheRS is not the only aaRS family member for which roles beyond charging tRNAs have been identified (Dolde et al., 2014). For instance, MetRS/MRS is capable of stimulating the rRNA synthesis (Ko et al., 2000), GlnRS/QRS can block the kinase activity of apoptosis signal-regulating kinase 1 (ASK1) (Ko et al., 2001) and a proteolytically processed form of YARS/TyrRS acts as a cytokine (Casas-Tinto et al., 2015; Greenberg et al., 2008). aaRSs are, however, also not the only protein family which evolved to carry out more than one function. In recent years it has become increasingly evident that many if not most proteins have evolved to carry out not only one, but two or more functions, providing interesting challenges to figure out, which of their activities are important for the individual functions of a protein (Dolde et al., 2014).
We found that elevated levels of α-PheRS promote cell proliferation in different cell types. In follicle cells, more cells were produced in the α-PheRS overexpressing clones compared to wild-type clones. In wing disc, more mitotic cells were detected in the α-PheRS overexpressing compartments. In the gut tissue, elevated α-PheRS(Cys) levels produced 2-5 times as many cells with intermediate size nuclei that stained positive for stem cell markers. Similarly, these guts also contained 5-8 times as many mitotic cells when α-PheRS(Cys) was overexpressed. Together, these results strongly suggest, that also in this situation the higher levels of α-PheRS induced over-proliferation of ISCs. These phenotypes are not only independent of the aaRS activity (Fig 1D-F), they also do not reflect a function in sensing the availability of its enzymatic substrate Phe and transmitting this information to the major growth controller, the TOR pathway (Fig S3). The proliferation activity of α-PheRS is therefore fundamentally different from the growth supporting activity of the aaRS members TrpRS or LeuRS (Adam et al., 2018; Bonfils et al., 2012; Han et al., 2012).
Overexpression of α-PheRS in the gut ISCs additionally interfered with cell fate decisions by driving ISCs to duplicate and to differentiate into EEs. The combination of these effects results in a “tumor-like” phenotype, that had been described as ISC/EE tumor phenotype that results from downregulation of Notch activity (Korzelius et al., 2014; Micchelli and Perrimon, 2006). Interestingly, mis-regulating proteins involved in EE fate specification, like downregulation of Tramtrack69 or upregulation of its adaptor protein Phyllopod, also results in this phenotype (Wang et al., 2015; Yin and Xi, 2018), suggesting that α-PheRS levels, through their activity on Notch signaling, induce the “tumor-like” phenotype in the gut tissue not only by increasing stem cell proliferation, but also by driving differentiating cells into an EE fate.
The gut phenotype caused by elevated levels of α-PheRS points to the importance of controlling and fine tuning these levels. We found that the JAK/STAT pathway has the capability of modulating cellular α-PheRS levels. Interestingly, JAK/STAT signaling, which stimulates progenitor cell differentiation during normal tissue homeostasis (Herrera and Bach, 2019), downregulates α-PheRS levels, which consequently allows higher Notch signaling to promote differentiation of the EB progenitor cells into ECs. Therefore, regulation of α-PheRS levels links the two signaling pathways and implicates α-PheRS not only in promoting cell proliferation, but also in regulating stem cell and tissue homeostasis by connecting the Notch to the JAK/STAT signaling pathway. Stat92E is a transcription factor, but none of the published reports on Stat92E targets lists the α-PheRS gene (Bina et al., 2010; Muller et al., 2005; Wang et al., 2013). A direct repression of α-PheRS gene expression by Stat92E appears therefore unlikely, even though it is also possible that this interaction escaped detection because it happens in too few cells of the analyzed cell types.
An interplay between the JAK/STAT and the Notch signaling pathways to maintain ISC homeostasis has been noted before in regenerating midguts. In this situation, damaged ECs release Upd cytokines to activate JAK/STAT signaling in ISCs and to promote ISC division (Jiang et al., 2009). This also stimulates Delta/Notch activity to promote EC differentiation (Jiang et al., 2009). Knocking down Stat92E reduced Notch signaling, and Notch target genes were downregulated when Stat92E activity was abolished in progenitor cells (Jiang et al., 2009). JAK/STAT signaling therefore feeds into the downstream Notch signaling pathway under these conditions, too. Our work revealed an unexpected missing link between the two signaling pathways in normal tissue homeostasis and it would therefore be interesting to also explore the function of α-PheRS in these pathways during tissue regeneration.
Relevance of moonlighting function of α-PheRS for tumor formation
Several aaRSs have come into the focus of cancer research (Kim et al., 2014). For instance, LeuRS senses the availability of the amino acid Leu and if these levels are sufficient to support growth, it signals a growth readiness to the key growth controller TORC1/mTORC1 (Bonfils et al., 2012; Han et al., 2012). Our results suggest that PheRS does not serve an analogous function as a Phe sensor (Suppl. Figure S3). Phe binding appears to be dispensable to activate proliferation and to repress Notch signaling because the α-PheRSCys mutant, in which two essential residues in the Phe binding pocket were replaced by Cys is unable to perform aminoacylation (presumably because it cannot bind Phe), but still able to induce the non-canonical activity.
Improper expression of PheRS was suspected long ago to promote carcinogenesis, but till now the mechanisms behind this effect remind unknown. Elevated levels of FARSA/CML33 (human α-PheRS) during the development of myeloid leukemia have been demonstrated to directly correlate with tumorigenic events (Sen et al., 1997). The GENT2 database published in 2019 describes also strong positive correlations between PheRS subunit levels and tumorigenic events in several tissues and cancers, including colon cancer, which mostly seems to originate from intestinal stem cells (ISCs) (Barker et al., 2009). Modelling the effect of elevated α-PheRS levels in Drosophila ISCs and CBs, we found that these levels lead to over-proliferation of cells with stem cell characteristics and to changes in cell fate, indicating that elevated α-PheRS levels can indeed be a risk factor for tumor formation.
Modeling the effects of elevated α-PheRS levels in ISCs revealed that hyperaccumulation of stem cells, a tumor risk, is mediated by high α-PheRS repressing Notch signaling. In mammals, Notch signaling is essential for maintaining the homeostasis of cell proliferation and differentiation (Qiao and Wong, 2009), similar to the function of Notch signaling in the Drosophila gut that is needed to prevent the induction of enteroendocrine tumors characterized by excessive EEs and ISCs in the adult midgut (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). Because in human, mis-regulation of Notch signaling in these processes has been suggested to trigger the development of colon cancer, Notch has been proposed as a molecular target for cancer therapy (Yin et al., 2010). The results presented here provide new and unexpected insights into the communication between two major signaling pathways involved in gut tumorigenesis and they suggest new opportunities to target these mechanisms.
Materials and Methods
Key Resources Table
Fly genetics and husbandry
All Drosophila melanogaster fly stocks were kept for long term storage at 18°C in glass or plastic vials on standard food with day/night (12h/12h) light cycles. All experiments were performed at 25°C unless specifically mentioned. A UAS-GFP element was added in the crosses of all rescue experiments to even out the effect of Gal4 by providing the same number of UAS constructs. Origins of all stocks are noted in the Key Resource Table.
DNA cloning and generation of transgenic flies
Sequence information was obtained from Flybase. All mutations and the addition of the Myc-tag to the N-terminus of α-PheRS were made by following the procedure of the QuickChange® Site-Directed Mutagenesis Kit (Stratagene). The genomic α-PheRS rescue construct (Myc::α-PheRS) codes for the entire coding region and for an additional Myc tag at the N-terminal end. In addition, it contains ∼ 1kb of up- and down-stream sequences and it was cloned into the pw+SNattB transformation vector (Koch et al., 2009; Lu et al., 2014). The α-PheRS and β-PheRS cDNAs were obtained by RT-PCR from mRNA isolated from 4-8 days old OreR flies (Lu et al., 2014). The Tyr412Cys and Phe438Cys mutations in the α-PheRS sequence were created by site directed mutagenesis. Like the wild-type cDNA, they were cloned into the pUASTattB transformation vector to generate the pUAS-α-PheRS and pUAS-α-PheRSCys. Before injecting these constructs into fly embryos, all plasmids were verified by sequencing (Microsynth AG, Switzerland). Transgenic flies were generated by applying the ϕ C31-based integration system with the stock (y w att2A[vas-ϕ]; +; attP-86F) (Bischof et al., 2007).
Western blotting
Protein was extracted from tissues, whole larvae, or flies using the lysis buffer. Protein lysates were separated by SDS-PAGE and transferred onto PVDF membranes (Milipore, US). The blocking was performed for 1h at room temperature (RT) with non-fat dry milk (5%) in TBST solution. Blots were probed first with primary antibodies (diluted in blocking buffer) overnight at 4°C and then with secondary antibodies (diluted in TBST) 1h at RT. The signal of the secondary antibody was detected by using the detect solution mixture (1:1) (ECL™ Prime Western Blotting System, GE Healthcare Life Science) and a luminescent detector (Amersham Imager 600, GE Healthcare Life Science). Origins and recipes of all buffers and reagents are noted in Key Resource Table.
Immunofluorescent staining and confocal microscopy
Guts were dissected from each female fly 3 days after eclosure, and a total of 10 guts were analyzed for each genotype. Dissections were performed in PBS 1X on ice and tissues were collected within maximum one hour. Fixation with 4% PFA in PBS-T 0.2% at RT was done for different durations depending on the different tissues: two hours (gut), 40 minutes (brain), 30 minutes (wing discs, ovary). Then the samples were blocked overnight with blocking buffer at 4°C. Primary antibodies (diluted in blocking buffer) were incubated with the samples for 8h at RT. The samples were rinsed 3 times and washed 3 times (20 minutes/wash) with PBST. Secondary antibodies (diluted in PBST) were incubated overnight at 4°C. The samples were then rinsed 3 times and washed 2 times (20 minutes/wash) with PBST. Hoechst 33258 (2.5 µg/ml) was added in PBST before the last washing step and the samples were mounted with Aqua/Poly Mount solution (Polysciences Inc., US). For the anti-Delta labeling, the samples were blocked for 3h at RT with blocking buffer. The primary anti-Delta antibody (1:10 v/v) was incubated with the samples overnight at 4°C and then the secondary antibody was incubated overnight at 4°C. Origins and diluted concentrations of all buffers and antibodies are noted in Key Resource Table.
Protein synthesis measurements using the ribopuromycylation method (RPM)
For puromycin labeling experiments, tissues were dissected in Schneider’s insect medium (Sigma, US) supplement with 10% fetal calf serum (FCS, Sigma, US) at 25°C. They were then incubated with Schneider’s insect medium containing puromycin (5 µg/ml, Sigma, US) and cycloheximine (CHX, 100 µg/ml, Sigma, US) for 2 hours at RT. Then the samples were fixed with 4% PFA in PBS-T 0.2% at RT and blocked overnight with blocking buffer at 4°C. Primary anti-Puromycin antibody (diluted in PBST) was incubated with the samples for 8h at RT. The samples were rinsed 3 times and washed 3 times (20 minutes/wash) with PBST. Secondary antibodies (diluted in PBST) were incubated overnight at 4°C. The samples were then rinsed 3 times and washed 2 times (20 minutes/wash) with PBST. Hoechst 33258 (2.5 µg/ml) was added in PBST before the last washing step and the samples were mounted with Aqua/Poly Mount solution (Polysciences Inc., US).
In vitro aminoacylation assay
Recombinant α-PheRS and β-PheRS proteins were expressed in the E. coli strain Rosetta (Novagen) and then purified (Moor et al., 2002). For this, the α-PheRS or α-PheRCys mutant cDNAs were cloned with His tags at the N-terminal end into the pET-28a plasmid expression vector (Novagen). Wild-type β-PheRS cDNAs were cloned into the pET LIC (2A-T) plasmid (Addgene). Then, His-α-PheRS or the His-α-PheRCys mutant and β-PheRS were co-expressed in the E. coli strain Rosetta with isopropylthiogalactoside (IPTG, 1mM) induction at 25 °C for 6 hours. Proteins were purified with Ni-NTA affinity resin (Qiagen). The aminoacylation assay protocol from Jiongming Lu was then followed (Lu et al., 2014). This assay was performed at 25 °C in a 100µl reaction mixture containing 50 mM Tris-HCl pH 7.5, 10 mM MgCl2, 4 mM ATP, 5 mM β-mercaptoethanol, 100 µg/ml BSA, 3 U/ml E. coli carrier tRNA, 5 µM [3H]-amino acid (L-Phe) and 1 µM tRNAPhe from brewer’s yeast (Sigma, US). In each experiment, a 15-µl aliquot was removed at four different incubation time points, spotted on a Whatman filter paper discs and washed three times with ice-cold 5% trichloroacetic acid and once with ice-cold ethanol. A blank paper disc without spotting and another with spotting the enzyme-free reaction were used for detecting background signals. After filter discs were dried, they were immersed into PPO Toluol (Sigma, US) solution in plastic bottles and the radioactivity was measured by scintillation counting.
Wing disc dissociation and FACS analysis
Wandering larvae derived from 2-4 hours egg collections were dissected in PBS during a maximal time of 30 minutes. Around twenty wing discs were incubated with gentle agitation at 29°C for around 2-hours in 500μl 10× Trypsin-EDTA supplemented with 50 μl 10×Hank’s Balanced Salt Solution (HBSS) (Sigma, US) and 10 μl Vybrant DyeCycle Ruby stain (Molecular Probes, US). Dissociated cells from wing discs were directly analyzed by FACS-Calibur flow cytometer (Becton Dickinson, US).
Drosophila tissue culture cells were harvested and fixed in 70% ethanol and stained with a staining solution containing 1mg/ml propidium iodide, 0.1% Triton and 10 mg/ml RNase A. The cells were then subjected to FACS-Calibur cytometry and data were analyzed with the FlowJo software.
Drosophila cell culture and RNAi treatment
The protocols for in vitro cell culture and RNAi treatment was described in the PhD thesis of Jiongming Lu (Lu, 2013). Drosophila Kc cells were incubated at 25°C in Schneider’s Drosophila medium supplemented with 10% heat-inactivated fetal calf serum (FCS) and 50 μg/ml Penicillin/Streptomycin. To induce RNAi knockdown in Drosophila cells, dsRNA treatment was performed (Clemens et al., 2000). dsRNAs around 500bp in length were generated with the RNAMaxxTM High Yield Transcription Kit (Agilent, US). Cells were diluted to a concentration of 106 cells/ml in serum-free medium, and dsRNA was added directly to the medium at a concentration of 15 μg/ml. The cells were incubated for 1 hour followed by addition of medium containing FCS. Then the cells were kept in the incubator and were harvested at different time points (1-5 days) after dsRNA treatment.
Clonal assay and twin spot data analysis
For twin spot tests, we used the Mosaic Analysis with a Repressible Cell Marker (MARCM) system. Twin spots were generated with the progenitor genotype hs-flp; tub-Gal4/UAS-β-PheRS; FRT82B, ubiGFP, UAS-α-PheRS(Cys)/FRT82B Tub-Gal80. In twin spots, the internal control clone was GFP-minus and the sister clone with the red signal generated by the antibody against the overexpressed protein. We induced the hs-FLP, FRT82B system at 37°C for 1h on the third day post-eclosure and dissected the animals 3 days post-induction. Confocal imaging detected non-green clones (without ubiGFP) and red clones (stained with Myc antibody-red) (Fig 9A).
In twin spots, cell numbers per clone were counted and the numbers of cell division per clone were calculated as log2(cell numbers per clone). This represents the logarithm of the cell numbers per clone to the base 2. The increase of cell proliferation (%) was analyzed by comparing the number of cell divisions of the two clones in the same twin spot.
Image acquisition and processing
Imaging was carried out with a Leica SP8 confocal laser scanning microscope equipped with a 405 nm diode laser, a 458, 476, 488, 496 and 514 nm Argon laser, a 561 nm diode pumped solid state laser and a 633 nm HeNe laser. Images were obtained with 20x dry and 63x oil-immersion objectives and 1024x1024 pixel format. Images were acquired using LAS X software. The images of the entire gut were obtained by imaging at the standard size and then merging maximal projections of Z-stacks with the Tiles Scan tool. Fluorescent intensity was determined from FIJI software.
Quantification of cell numbers per posterior midgut
Z stack images through the width of the posterior midgut were acquired along the length of the posterior midgut from the R4a compartment to midgut-hindgut junction. Maximum projections of each Z stack were obtained, and the total number of each cell type was counted manually and exported to Microsoft Excel and GraphPad Prism for further statistical analysis.
Quantification and statistical analysis
For quantifications of all experiments, n represents the number of independent biological samples analyzed (the number of guts, the number of wing disc, the number of twin spots), error bars represent standard deviation (SD). Statistical significance was determined using the t-test or ANOVA as noted in the figure legends. They were expressed as P values. (*) denotes p < 0.05, (**) denotes p < 0.01, (***) denotes p < 0.001, (****) denotes p < 0.0001. (ns) denotes values whose difference was not significant.
Author contributions
T.H., J.L. and B.S. conceived the ideas and designed the experiments. T.H. conducted most experiments and performed the analysis of the results. J.L. performed the loss-of-function experiments of PheRS and the mTOR signaling tests, the adult wing measurements and the FACS analysis, including analyzing their data. T.H., J.L. and B.S. wrote the manuscript.
Conflict of interests
The authors declare that they have no conflict of interest.
Supplementary Figure S2
Analyzing the size of dissociated larval wing disc cells by FACS revealed that the cells from the posterior compartment (GFP-positive compartment), where both PheRS subunits were overexpressed, were on average larger than the ones that overexpressed only GFP or only α-PheRS (A). In contrast, all control cells from the anterior (GFP-negative) compartment of these three lines were of similar, smaller size (A). Cell numbers, on the other hand, did not significantly change upon overexpression of both PheRS subunits (B). We conclude that in larval wing discs with their organ size control mechanism, PheRS overexpression causes primarily an increase in cell size. On the other hand, single overexpression of α-PheRS increases the mitotic index of wing discs (Fig 1A-C).
Supplementary Figure S3
Amino acid deprivation downregulates phosphorylation of dS6K in Kc cells, and subsequent stimulation with amino acids can restore this phosphorylation (Kim et al., 2008). The Rag complex is part of a nutrient sensor pathway, and its knockdown prevents the TORC1 complex from sensing the availability of amino acids (Kim et al., 2008; Sancak et al., 2008). In contrast, when β-PheRS was knocked down, amino acids were still able to induce phosphorylation of dS6K to similar levels as in the control. In this case, it did not matter whether all amino acids were added back (A) or only L-Phe (B). These results therefore suggest that PheRS does not serve as an amino acid sensor upstream of the TORC1 complex, although we cannot rule out that the knockdown was not sufficient to induce this effect.
Acknowledgements
We thank Peter Nagy, Hugo Stocker, Albena Jordanova, Erik Storkebaum and the Bloomington Stock Center for fly stocks. We are also grateful to Mark Safro for suggesting mutations that disrupt the phenylalanine binding site of α-PheRS. This work was supported by the Novartis Foundation for Medical-Biological Research (#18A050), the Swiss National Fund (project grant 31003A_173188) and the University/Canton of Bern.