J-Perk 6.05 serial key or number

1. 1.Glaxo Wellcome., Inc.Research Triangle Park

• DOIhttps://doi.org/10.1385/0896035174
• Copyright InformationHumana Press 1999
• Publisher NameHumana Press
• eBook PackagesSpringer Protocols
• Print ISBN978-0-89603-517-1
• Online ISBN978-1-59259-605-8
• Series Print ISSN1543-1894
• Series Online ISSN1940-6037
• Buy this book on publisher's site

Cellular stress response contributes to epithelial defense in adaptation to environment changes. Galectins play a pivotal role in the regulation of this response in malignant cells. However, precise underlying mechanisms are largely unknown. Here we demonstrate that Galectin-3, a pro and anti-apoptotic lectin, is required for setting up a correct cellular response to stress by orchestrating several effects. First, Galectin-3 constitutes a key post-transcriptional regulator of stress-related mRNA regulons coordinating the cell metabolism, the mTORC1 complex or the unfolded protein response (UPR). Moreover, we demonstrated the presence of Galectin-3 with mitochondria-associated membranes (MAM), and its interaction with proteins located at the ER or mitochondrial membranes. There Galectin-3 prevents the activation and recruitment at the mitochondria of the regulator of mitochondria fission DRP-1. Accordingly, loss of Galectin-3 impairs mitochondrial morphology, with more fragmented and round mitochondria, and dynamics both in normal and cancer epithelial cells in basal conditions. Importantly, Galectin-3 deficient cells also display changes of the activity of the mitochondrial respiratory chain complexes, of the mTORC1/S6RP/4EBP1 translation pathway and reactive oxygen species levels. Regarding the ER, Galectin-3 did not modify the activities of the 3 branches of the UPR in basal conditions. However, Galectin-3 favours an adaptative UPR following ER stress induction by Thapsigargin treatment. Altogether, at the ER-mitochondria interface, Galectin-3 coordinates the functioning of the ER and mitochondria, preserves the integrity of mitochondrial network and modulates the ER stress response.

Eukaryotic cells are partitioned into discrete organelles endowed with specific functions. For example, protein synthesis, folding, secretion and degradation take place at the endoplasmic reticulum (ER), whereas cellular bioenergetics occurs at the mitochondria. Cell homeostasis depends on the ability of these distinct organelles to communicate and to exchange cargos. Among them, the ER appears especially important since it can develop close contacts with most organelles, notably the mitochondria¹. Interestingly, the ER-mitochondria interactions which are commonly referred to as mitochondria-associated membranes (MAM) regulate important cellular processes such as mitochondrial biogenesis and metabolism, calcium signalling or lipid biosynthesis^2,3. When the protein synthetic load overcomes the ER folding capacities, a state called ER stress occurs. To restore ER homeostasis, a network of transduction pathways named unfolded protein response (UPR) is activated⁴. Collectively, these pathways act to reduce protein synthesis and to increase ER folding and ER-associated degradation (ERAD) capacities. However, when this adaptative response fails the UPR evolves towards apoptosis.

Galectin-3 is a β-galactoside-binding lectin, encoded by LGALS3 gene in humans, which contains a C-terminal carbohydrate recognition domain (CRD) responsible for interactions with glycolipids or glycoproteins and a low complexity domain which allows interactions with the CRD and other partners^5,6. Moreover, despite the absence of a canonical RNA-binding domain, Galectin-3 is a non-classic RNA-binding protein (RBP) able to stabilise mucin MUC4 mRNAs in cancer cells⁷. Galectin-3 is highly expressed by epithelial cells and plays important roles in the organisation of renal and intestinal cells. Although Galectin-3-KO mice are viable in controlled conditions, loss of Galectin-3 leads to morphological abnormalities of the epithelial cells as well as perturbation of the biosynthetic pathway^8,9,10. Galectin-3 is a soluble protein which is synthesised on free ribosomes and thus bypasses the classical ER-Golgi pathway for its secretion in the extracellular medium. Indeed, premature binding of Galectin-3 with its ligands which are major components of the ER lumen would cause aggregation and perturb the secretory pathway^11,12. While being synthesised in the cytosol, Galectin-3 associates with various organelles, such as carrier vesicles or endosomes¹³. At the mitochondria level, Galectin-3 prevents the cytochrome-c release and ensures mitochondrial integrity^14,15. However, it is currently unknown whether these mitochondrial effects depend on Galectin-3 ability to modulate mitochondria-ER interactions in epithelial cells.

In the present study we first aimed to obtain a global view of the post-transcriptional regulatory action of Galectin-3 in epithelial cells. To this end, we combined whole transcriptome stability analysis with mRNA and protein quantification. We showed that Galectin-3 regulates the stability of subsets of mRNAs which share similar functions notably cell metabolism, cell death and stress response pathways. By coupling imaging and biochemical approaches, we showed that Galectin-3 localises at the ER-mitochondria interface where it preserves the integrity of the mitochondrial network and modulates the cellular bioenergetics and the UPR.

Gal-3 regulates the half-life of subsets of mRNAs with shared functions

We first aimed to obtain a global view of the action of Galectin-3 as a post-transcriptional regulator in epithelial cells. For that purpose, we used two models deriving from the human pancreatic cancer cell line T3M-4, control Sc cells expressing high levels of Galectin-3 and a representative mutant clone (Sh1 called Sh cells thereafter) where Galectin-3 expression was stably knocked-down by shRNA (Fig. S1). Using actinomycin D and RNA-Seq, we performed a whole-transcriptome stability analysis. We built the decay curve for 9444 transcripts in both models using the most appropriate kinetic model. In all, 3356 transcripts with a decay curve characterised by a r² ≥ 0.9 were selected for further analysis. Rapidly decaying mRNAs (t_1/2 < 3.0 h) were excluded as the experimental time-points did not allow reliable calculation of their half-lives. Therefore, the total number of RNAs analysable was 2275 (Fig. 1a). A total of 440 mRNAs showed a ≥2-fold difference in their half-life between Sc and Sh cells (Table S1). The vast majority (385 mRNAs) were stabilised by Galectin-3 whereas only 55 mRNAs were destabilised by Galectin-3 (Fig. 1a, b and Table S1). To further identify the biological processes and the organelles involved, the set of mRNAs whose stability was modified by Galectin-3 (n = 440) was submitted to an analysis by the Gene Set Enrichment Analysis (GSEA) tool (Fig. 1c, d). Focusing on cellular component, the mitochondria and the ER were significantly enriched in proteins whose mRNAs are post-transcriptionally regulated by Galectin-3 (Fig. 1c, Table S2). Using Hallmark as a reference collection, 13 biological pathways were significantly enriched in our data set, several previously unrelated to Galectin-3 such as metabolic pathways (i.e. glycolysis, fatty acid metabolism) and the UPR (Fig. 1d and Table S2). Regarding UPR, twelve mRNAs stabilised by Gal-3 were classified in this category by GSEA (in black). Twelve additional mRNAs (in blue) were assigned to UPR based on the Gene Ontology (GO) annotation of the gene product as determined with AmiGO (Fig. 1d). To check whether changes of the mRNA half-life influence mRNA levels, we evaluated by qPCR the steady state mRNA levels of 7 mRNA from the UPR category (Fig. 1d, underlined) and stabilised by Galectin-3. All but one mRNA showed a significant reduction of expression in Sh cells by comparison with Sc cells (Fig. 1e). Moreover, we studied protein expression of four mRNA stabilised by Galectin-3, namely PARN, XPOT, USP14 and SEL1L in basal state. Our results showed that the protein levels of USP14 and SEL1L, both involved in ERAD, were significantly higher in Sh cells whereas a similar trend existed for PARN and XPOT (Fig. 1f). Thus Galectin-3, as a non-conventional RBP, modulates selectively the half-life, levels and translation of mRNAs whose encoded proteins share similar biological functions (RNA regulons).

Localisation of Galectin-3 at the ER-mitochondria interface

Since Galectin-3 post-transcriptionally regulates several genes involved in mitochondria and/or ER functioning, we next wondered whether Galectin-3 could localise at the ER-mitochondria interface. Using high-resolution fluorescence microscopy, KDEL as a proxy of the ER network and Mitotracker to stain the mitochondria (Fig. 2a), we showed that, although the ER and mitochondria display by themselves 12% of contacts, ~13% of the ER-mitochondria interfaces associate with Galectin-3 in pancreatic Sc cells (Fig. 2b). Conversely, 4% of total Galectin-3 associates with this area (as expected for a highly expressed lectin which associates with diverse subcellular compartments^13,16). In addition, live microscopy using Mitotracker and Gal-3 dsRed construct validated the dynamic association between Galectin-3 and the mitochondria over time (Movie S1). Immunogold staining of human intestinal Caco2 cells coupled with transmission electron microscopy (TEM) revealed that Galectin-3 does associate with the ER complex. However, this association is transient, and specifically occurs at contact sites between the ER and the mitochondria (Fig. 2c, yellow circles and arrows). Finally, we performed density gradient centrifugations to collect subcellular fractions enriched in mitochondria, MAMs and ER, according to a reference method. We confirmed that a pool of intra-cellular Galectin-3 is present in MAM-enriched fraction in both Caco2 and pancreatic Sc cells (Fig. 2d). The presence of Galectin-3 in the ER fraction could be explained by its possible contamination by lysosomes. Interestingly, among the interacting partners of Galectin-3 identified in Caco-2 cells by immunoprecipitation (IP) coupled with mass spectrometry analysis, seven (≈4% of the whole) are exclusively located at the ER membrane such as TBL2, a regulator of ER stress response, and 13 in the mitochondria such ATP5J (Table S3 and Fig. S2). To conclude, by combining several approaches we show that Galectin-3 is found at MAMs thus suggesting a dedicated function of Galectin-3 at that level.

Impact of Galectin-3 on mitochondria morphology, dynamics and bioenergetics

The ER-mitochondria interface constitutes a key hub for the ER stress response and mitochondria bioenergetics^17,18. We then hypothesised that Galectin-3 may be instrumental there for cell adaptation to stress and/or mitochondrial homeostasis. First, we studied the impact of Galectin-3 on the morphology and behaviour of mitochondria. Ultrastructural analyses performed in WT and Lgals3^−/− mouse enterocytes (Fig. 2e) showed that loss of Galectin-3 provokes the formation of enlarged and swollen mitochondria whereas in wild-type cells mitochondria display the classical “stick shape”. As expected, image analysis confirmed an increased maximum diameter in Galectin-3 deficient versus control mouse enterocytes (Fig. 2f). Similarly, ultrastructural analysis of the mitochondrial network in Sh cells revealed irregular mitochondrial shape in comparison with controls Sc cells (Fig. 2g). In addition, numerous degradative compartments appear in close proximity of mitochondria in Sh cells. We concluded that Galectin-3 is required for maintenance of typical mitochondrial morphology in epithelial cells. Second, we analysed the architecture of the mitochondrial network (Fig. 3a). In dense cultures, analysis of the images with the MiNA toolset¹⁹ enabled to calculate the values of mitochondrial network 2D parameters. The mitochondrial footprint (Fig. 3a, b), the number of branches per network and the summed branch length (Fig S3a) were significantly decreased in Sh cells versus control Sc cells. These results show that, in the absence of Galectin-3, the mitochondrial network exhibits a less tubular shape but instead mitochondrial fragmentation (Fig. 3a, d, e). Alterations of the basal ER stress level in Sh cells could be responsible for this phenotype. Along this line, we induced acute ER stress response by treating Sc and Sh cells with 250 nM Thapsigargin alone, or in combination with either GSK2606414, a potent and selective inhibitor of PERK, or ISRIB, a PERK-independent stabiliser of eIF2α dephosphorylation. Thapsigargin in Sc cells induced a significant decrease of mitochondrial footprint (Fig. 3b), of the number of branches per network and of the summed branch length (Fig. S3), leading to mitochondrial characteristics very similar to those of Sh cells. As expected, combined treatment with GSK2606414 or ISRIB partially reversed mitochondrial morphology in Sc cells. On the contrary, in the absence of Galectin-3, treatment with Thapsigargin very negligibly altered the mitochondrial footprint and branching. Moreover, combination with GSK2606414 or ISRIB had no significant effect compared to Thapsigargin treatment alone, and did not revert mitochondrial morphology, indicating that mitochondrial fragmentation in Sh cells is independent of UPR. Hence, the loss of Galectin-3 silencing may provoke a perturbation of the mitochondrial dynamics, most probably in relation with fission/fusion events. Therefore, we analysed the behaviour of DRP-1 which is the central modulator of mitochondrial fission²⁰. Phosphorylation of Drp-1 at Ser637 prevents its translocation from the cytosol to the mitochondria and its fission activity. By contrast, pSer616 Drp-1 stimulates mitochondrial fission²¹. Here we observed by immunofluorescence that pSer637 Drp-1 expression was higher in Sc than Sh cells, a trend further confirmed by western blotting analysis of mitochondrial extracts (Fig. 3d, e). By contrast, pSer616 Drp-1 was highly expressed in Sh cells and frequently co-localised with Tom20, a mitochondrial marker. Total Drp-1 level was also significantly increased in Sh cells (Fig. S3b). Altogether, these data testify a mitochondrial pro-fission phenotype in the absence of Galectin-3. Moreover, to test mitochondrial plasticity, we performed live cell imaging and mitochondrial motility was assessed by particle-tracking analysis of Mitotracker-labelled mitochondria in Sc (Movie S2) and Sh cells (Movie S3). Whatever the parameter used (i.e. speed or distance), the motility of mitochondria was significantly decreased in absence of Galectin-3 (Fig. 4a, b). Altogether, these results reveal that Galectin-3 influences the mitochondrial dynamics/homeostasis by inhibiting the fission process and promoting the motility instead.

To determine whether morphological abnormalities may impinge on mitochondrial activity, the cell respiration was studied by oxygraphy using O2k (OROBOROS instruments). Results showed that the activities of mitochondrial Complex I (NADH dehydrogenase) and Complex IV (cytochrome C oxydase) were significantly higher in Sh versus control pancreatic Sc cells (Fig. 5a). The activities of Complex II and III did not differ between Sh and control Sc cells. The electron transfer system (ETS) measured after addition of FCCP, an uncoupling agent, was significantly higher in Sh than control Sc cells (Fig. 5a). By contrast no significant difference was observed for Complex V nor ETS measured in coupled conditions (C ETS). In intestinal cells, no significant differences in the activities of Complexes I–V and ETS were observed between H3 and NT control cells, probably because of the high SD values. However, despite the versatility of Caco2 cells, Galectin-3 depleted cells presented a reproducible and significant increase in C ETS when compared to control cells (p = 0.0169).

Since oxidative phosphorylation (OXPHOS) is the main contributor to ATP cellular production and occurs downstream of the glycolysis, we hypothesised that this inhibitory effect of Galectin-3 may induce a cellular metabolic stress. We therefore measured extracellular lactate levels as a proxy of the glycolytic activity. Lactate levels were higher in Sc versus Sh pancreatic cells (Fig. 5b, p < 0.05), whereas no difference occurred between Galectin-3-silenced Caco2 cells and parental cells. However, the ADP/ATP ratio, which is inversely related to the cell energy status, was not significantly lower in Sh than control Sc cells. Thus, in the presence of Galectin-3 the mitochondrial OXPHOS is less efficient but no metabolic stress is observed. Leakage of electrons from the mitochondrial electron transport chain results in the generation of reactive oxygen species (ROS) mainly at the level of complexes I and III²². By flow cytometric analysis of MitoSOX® Red-stained populations, we herein showed that the number of cells with positive mitochondria was not different between Sc and Sh cells (Figs. 5c and S4). However, there was a higher percentage of high ROS cells in Sh than in control Sc cells (44% vs 16.5%, p < 0.0001) in agreement with a higher OXPHOS activity in the mitochondria of Galectin-3 silenced cells.

To explain these differences in mitochondria dynamics and bioenergetics, we wondered whether Galectin-3 might affect the expression of components of the respiratory chain post-transcriptionally. In this respect, we identified two species of interest among the mRNAs destabilised by Galectin-3 (Table S1). First, PDSS2 which encodes DLP1, a protein involved in the biosynthesis of coenzyme Q which promotes electron transfer between complexes II and III²³, and GUF1 alias EF4 which encodes a protein of the mitochondrial inner membrane important for complex IV organisation and activity²⁴. By western blotting (Fig. S5), we showed that Galectin-3 did not influence the expression of DLP1 or GUF1. The same was observed for ISCU, a scaffold protein involved in the assembly of Fe–S cluster found in complexes I, II and III. However, among the interacting partners of Galectin-3 identified by co-immunoprecipitation we did find ICSU, and also other members of mitochondrial complexes such as UQCRC2 (Complex III) or the F0 complex of ATP Synthase (Table S3).

Gal-3 selectively controls mRNA translation through mTORC1

Since mTORC1 is a global sensor of the cell nutrient status we wondered whether Galectin-3 expression may influence mTORC1 activity due to its negative impact on OXPHOS. Addition of fetal calf serum (FCS), an activator of mTORC1, in serum-starved cells failed to induce S6RP and 4EBP1 phosphorylation in control cells by contrast with Sh cells (Fig. 6a). In agreement, a trend towards higher levels of phosphorylated mTOR at Ser2448 was observed in Sh cells vs control cells but the phospho-mTOR/mTOR ratio was not different between Sh and Sc, probably because a decrease of total mTOR occurred in Sc cells (Figs. 6a and S6). No difference was observed for the AMPK pathway between Sc and Sh cells, in accordance with the absence of metabolic stress reported earlier (Fig. 5b). Thus the presence of Galectin-3 decreases S6RP and 4EBP1 activity in part by reducing mTORC1 activation. Since phosphorylation of both S6RP and 4EBP favours mRNA translation initiation, Galectin-3 inhibitory effect may induce a blockage of mRNA translation in pancreatic cancer cells in usual culture conditions (with FCS). To validate this point, we prepared protein extracts and total RNA simultaneously and measured the protein levels by western blot and mRNA levels by qPCR of four genes representative of the mTORC1 and fatty acid metabolism GO categories whose mRNAs are stabilised by Galectin-3. Whereas GBE1 and SLC2A3 mRNA levels were higher in Sc cells than in Sh cells, the corresponding protein levels were unchanged or decreased in Sc versus Sh cells, respectively (Fig. 6b). Similarly, ACOT2 protein levels were decreased in Sc mitochondrial extracts whereas its mRNA levels were similar between Sc and Sh cells. Thus, the differential expression of transcript and protein observed for GBE1, ACOT2 and SLC2A3 but also for USP14 and XPOT (Fig. 1f) suggests that, in presence of Galectin-3, the translation of a fraction of these 5 mRNAs is blocked (Fig. 6c). A second category of mRNAs is represented by ETFDH whose mRNA and protein levels were unaffected by Galectin-3 expression (Fig. 6). In conclusion, Galectin-3 appears to reprogram selectively the translatome of pancreatic cancer cells in basal conditions through the control of S6RP and 4EBP1 activity.

Gal-3 promotes an early transactivation of ATF6- and Xbp-1s target genes following ER stress induction

In mammalian cells, three canonical ER resident transmembrane proteins (IRE1, PERK and ATF6) act as ER stress sensors and activate three pathways (Fig. 7a). mRNA expression levels of 84 key UPR genes were then monitored by PCR arrays, at baseline (T0) and 1, 4 and 16 h following ER stress induction by treatment with 250 nM Thapsigargin. Levels of 31 mRNAs showed significant variations, among which 26 were significantly increased at early phases of cell response to Thapsigargin treatment whereas 5 were diminished in control Sc cells versus Sh cells respectively (Fig. 7b). Most of these mRNA inductions in Sc cells occurred at steady state (38.5%) or after 1 h of Thapsigargin treatment (61.5%). Based on promoter structure analysis²⁵, the 31 mRNAs which showed differential expression between Sc and Sh cells were classified according to the specific contribution of ATF4, ATF6(N) and Xbp-1s transcription factors for their expression (Fig. 7b). Interestingly, seven mRNAs correspond to ATF6(N)-target genes, three mRNAs to Xbp-1s-target genes, three mRNAs to ATF6(N) and Xbp1-s target genes, whereas only one mRNA corresponds to ATF4-target genes. Next we decided to study the impact of Galectin-3 on the three UPR branches at the protein level. As soon as 1 h of Thapsigargin treatment, the ratio cleaved ATF6/full length AT6 protein was significantly increased in control Sc cells, demonstrating an engagement of the ATF6 arm (Fig. 7c). Hence, we asked whether this could translate into early transactivation of specific ATF6 targets. Indeed, we observed that 1 h after Thapsigargin treatment ATF6B and MBTPS1 mRNA levels (Figs. 7c and S7) were significantly increased in Sc versus Sh cells. Regarding the IRE1 branch (Fig. 7d), Xbp-1s protein expression was increased in control Sc cells as early as 1 h after Thapsigargin addition (p < 0.05 versus baseline) whereas in Sh cells activation was delayed. Maximal induction of Xbp-1s occurred at 4 h time course in both cell lines (Figs. 7d and S7). Accordingly, some XBP-1s-dependent targets such as ERP44, EDEM1 (Fig. 7d) or HERPUD1 (Fig. S7) were also significantly up-regulated in control Sc cells 1 h after Thapsigargin treatment. Finally, concerning the PERK arm (Fig. 7e), phosphorylation of eif2α was delayed in control Sc cells as was ATF4 protein expression. In fact, ATF4 protein levels were significantly higher as early as 1 h after Thapsigargin addition only in Sh cells (p < 0.05 versus baseline, Fig. S5). Since ATF4 mRNA levels did not vary significantly during Thapsigargin treatment, ATF4 protein expression results from a phospho-eif2α-dependent translation reprogramming. Taken together, these data show that Galectin-3 favours an early activation of the ATF6 and IRE/Xbp1s arms of the UPR resulting in an increase of the ER protein folding (ERP44, CALR) as well as ERAD pathway (HERPUD1) capacities. Moreover, Galectin-3 also delays ATF4 expression during an experimental ER stress. Finally, we studied the influence of Galectin-3 on cell viability during a prolonged Thapsigargin treatment (>16 h). By live and dead assay, we demonstrated that the mortality of control Sc cells was significantly higher than those of Sh cells at 24 and 48 h of the treatment (Fig. 7f).