Metastatic lymph node 51 (MLN51, also known as CASC3) is a core component of the exon junction complex (EJC), which is loaded onto spliced mRNAs and plays an essential role in determining their fate. Unlike the three other EJC core components [eIF4AIII, Magoh and Y14 (also known as RBM8A)], MLN51 is mainly located in the cytoplasm, where it plays a key role in the assembly of stress granules. In this study, we further investigated the cytoplasmic role of MLN51. We show that MLN51 is a new component of processing bodies (P-bodies). When overexpressed, MLN51 localizes in novel small cytoplasmic foci. These contain RNA, show directed movements and are distinct from stress granules and P-bodies. The appearance of these foci correlates with the process of P-body disassembly. A similar reduction in P-body count is also observed in human HER2-positive (HER2+) breast cancer cells overexpressing MLN51. This suggests that P-body disassembly and subsequent mRNA deregulation might correlate with cancer progression.
MLN51 (metastatic lymph node 51, also known as CASC3) is a protein involved in mRNA metabolism through its association with the exon junction complex (EJC). The EJC is a macromolecular complex that is deposited by the splicing machinery onto mRNAs at the exon–exon junction. EJC binding is a crucial step in determining the subsequent fate of mRNA, as the EJC has been proposed to regulate pre-mRNA splicing, cytoplasmic export, localization, translation and transcript stability (Degot et al., 2004; Nott et al., 2004; Palacios et al., 2004; Gehring et al., 2005; Tange et al., 2005). The N-terminal region of MLN51 harbors the speckle localizer and RNA-binding module (SELOR), a unique RNA-binding domain that triggers the incorporation of MLN51 into the exon junction complex core (Degot et al., 2004; Ballut et al., 2005). The C-terminal moiety of MLN51 contains a nuclear export signal that allows the protein to shuttle between the nucleus and the cytoplasm and, interestingly, it is this particular region that is essential for the recruitment of MLN51 to cytoplasmic stress granules (Baguet et al., 2007).
Stress granules self-assemble in response to a wide variety of environmental stresses that can cause general translational arrest (Anderson and Kedersha, 2006). Several factors involved in translation, mRNA decay or mRNA silencing have been found to localize in stress granules (Kedersha and Anderson, 2007), leading to the description of stress granules as cytoplasmic compartments involved in the remodeling of messenger ribonucleoprotein particles (Mollet et al., 2008). They share some components with other non-membranous cytoplasmic structures called processing bodies (P-bodies) that are involved in mRNA storage, translational repression and decay (Eystathioy et al., 2002; Sheth and Parker, 2003; Cougot et al., 2004; Pillai et al., 2005; Eulalio et al., 2007; Parker and Sheth, 2007). Recently, we deciphered the structural organization of P-bodies and showed that they are organized into two compartments – a dense central core containing mRNA decay factors and a peripheral shell enriched with polysomes and translational activators (Cougot et al., 2012; Cougot et al., 2013). Overall, the presence of these two cytoplasmic substructures fits the ‘mRNA cycle’ model, according to which cytoplasmic mRNAs cycle between active polysomes, P-bodies and stress granules (Balagopal and Parker, 2009).
Because MLN51 was recently shown to be a translational activator and to be present in active polysomes (Chazal et al., 2013), we investigated the involvement of MLN51 in the mRNA cycle and explored the role of this protein in stress granules and P-bodies in HeLa and malignant breast cancer cells. In the present study, we show that endogenous MLN51 is a novel component of P-bodies. When overexpressed, it accumulates in two types of cytoplasmic foci – in stress granules and in a novel mRNA-containing granular structure that we named SMIGs (small MLN51-induced granules). These newly described cytoplasmic bodies contain RNAs and show directed movements. Moreover, MLN51 overexpression leads to microtubule-dependent P-body disassembly. Interestingly, a similar reduction in P-body count and MLN51 localization in small granules is also observed in human HER2+ breast cancer cells overexpressing MLN51, suggesting that P-body disassembly and subsequent mRNA deregulation might be linked to certain cancers.
Endogenous MLN51 is a novel component of P-bodies
To better characterize the cytoplasmic pattern of MLN51 distribution, we raised polyclonal antibodies directed against the N-terminal part of the protein (Degot et al., 2002). MLN51 was first detected by immunofluorescence in HeLa cells, which were also immunolabelled with antibodies against Ge1 (also known as EDC4) or human Dcp1a that allow detection of P-bodies (Kedersha and Anderson, 2007) (Fig. 1A,B). We observed that endogenous MLN51 accumulated in cytoplasmic foci that colocalized with both P-bodies markers. MLN51 is, in fact, the first member of the EJC core to be detected in P-bodies. To more precisely assess its localization, we performed immunoelectron microscopy using the same antibodies (Fig. 1C). MLN51-labeling gold particles were enriched by a factor of 48.47 in the P-bodies compared to the surrounding cytoplasm – 1990.54±1354.16 gold particles/µm2 (mean±s.d.) inside the P-bodies (n = 11) and 41.07±17.04 gold particles/µm2 outside. Quantification of the gold-labeling in the P-bodies was performed in an internal P-body area that was calculated by a 50% downscaling of the external contour added to the peripheral zone between it and the entire electron-dense body (Cougot et al., 2012). The peripheral and internal areas correspond to 75% and 25% of the total area, respectively. About 75% of the MLN51-labeling gold particles accumulated at the periphery of mammalian P-bodies (604 of the 806 gold-particles counted on 11 different hDcp1a gold-labeled P-bodies). Collectively, these results show that MLN51 is distributed equally throughout the granule.
Overexpression of MLN51 leads to P-body disassembly
We next analyzed the effects of MLN51 overexpression on P-bodies. To do so, CFP–MLN51 was expressed in HeLa cells, and we performed co-labeling of P-bodies using an anti-Dcp1a antibody (Fig. 2A). Surprisingly, MLN51-overexpressing cells showed a reduced number of P-bodies compared with that of untransfected cells, even though the levels of endogenous Dcp1a are not affected by MLN51 overexpression (Fig. 2B). P-body disassembly following MLN51 overexpression was confirmed when using various P-body markers, such as Dcp1, Ge1, rck/p54 or Xrn1 (Fig. 2C). We performed quantitative analysis and observed a 1.9-fold decrease in the number of cells containing at least one P-body (29 out of 54 transfected cells with at least one P-body, Fig. 3C,D). The average number of P-bodies per cell was decreased by a factor of 3.5 (2.66 P-bodies per cell in untransfected cells versus 0.76 in cells overexpressing CFP–MLN51), and the distribution of the number of P-bodies per cell showed a shift towards cells having fewer P-bodies, with a strong increase in the number of cells having zero or one. In fact, after MLN51 overexpression, 88% of the cells had less than one P-body, whereas this was the case in only 5.7% of untransfected cells (Fig. 2D).
P-body disassembly is not linked to the role of MLN51 in the EJC
We wanted to explore whether the effect of MLN51 overexpression on P-bodies is linked to the role of MLN51 in the EJC. MLN51 is composed of two domains (Fig. 3A): (1) the N-terminal domain contains a nuclear localization signal and is involved in the recruitment of MLN51 to the EJC core through the SELOR domain (Degot et al., 2004; Ballut et al., 2005) and (2) the C-terminal domain contains a nuclear export signal and plays a role in the recruitment of the protein to cytoplasmic stress granules (Baguet et al., 2007). First, we confirmed by western blot experiments that overexpression of MLN51 fragments had no effect on the level of expression of endogenous p54 and Dcp1a (Fig. 3B). Then, we observed that overexpression of the MLN51 N-terminus or SELOR did not have a significant effect on P-body number, irrespective of the P-body marker examined [hDcp1 (Fig. 3C) or p54 (Fig. 3D)]. The overexpression of the MLN51 C-terminal fragment induced a similar phenotype to that of the full-length protein (Fig. 3C,D). Taken together, these results suggest that the effect of MLN51 on P-body number is not linked to its association with the EJC. To confirm this, we overexpressed a mutant of MLN51 (H220A/D221A, MLN51HD) that has previously been shown to prevent MLN51 incorporation into EJCs both in vitro and in vivo (Ballut et al., 2005; Daguenet et al., 2012). As shown in Fig. 3C,D, overexpression of this mutant led to the disassembly of P-bodies (∼40% of the cells overexpressing MLN51HD contained at least one PB). Finally, we overexpressed two other core EJC components, eIF4A3 and Y14, and no significant effect on P-bodies was noticed (supplementary material Fig. S1). Similar observations were made with other P-body markers (Ge1, Xrn1).
Overexpression of MLN51 leads to its accumulation in SMIGs
Unlike the three other core EJC components (eIF4AIII, Magoh and Y14), MLN51 is primarily found in the cytoplasm, and we observed that the protein is present in P-bodies. As previously reported (Baguet et al., 2007), we confirmed the distribution of overexpressed MLN51 into three cytoplasmic fractions – diffuse (18.5%), small foci (77.8%) and larger foci (3.7%) (Fig. 2; Fig. 4C). Given that MLN51 is a stress granule component and participates in stress granule assembly, we wondered whether these foci could be related to stress granules. To better identify the cytoplasmic structures, we transiently expressed a CFP-tagged version of MLN51 in HeLa cells, then performed a series of co-stainings using antibodies against three stress granule markers – TIA1, FMRP and PABP (Buchan and Parker, 2009). Interestingly, we observed that the small MLN51 foci were negative for TIA1, FMRP and PABP markers (Fig. 4). Given that the larger foci did colocalize with all of these stress granule components (supplementary material Fig. S2A–C), this suggests that only the larger foci are stress-granule-related; the small foci correspond to unknown granules that we named ‘small MLN51-induced granules (SMIGs). It is important to note that the appearance of SMIGs is associated only with the expression of full-length MLN51, whereas the larger foci can also be observed following the overexpression of the MLN51 C-terminal domain.
SMIGs are mobile RNA granules
To further characterize SMIGs, we analyzed whether these granules contained RNA. Poly(A)+ mRNAs were detected by in situ hybridization using Cy5-labeled oligo d(T) probes in HeLa cells overexpressing CFP–MLN51. Triple fluorescence labeling experiments were then performed in order to determine the colocalization of MLN51, poly(A)+ and ATXN2, a component of stress granules. We found that poly(A)+ mRNAs actually colocalized with MLN51 and ATXN2 in stress granules (Fig. 5A) and only with MLN51 in SMIGs (Fig. 5B). Co-staining experiments of SMIGs with antibodies directed against the nuclear form of PABP and against the nuclear cap-binding protein CBP80 (also known as NCBP1) were performed, in order to determine whether mRNAs in SMIGs have actually been translated. However, SMIGs were negative for these both markers (supplementary material Fig. S3).
Thus, one possibility was that a SMIG is a precursor to a stress granule. To address this, we first studied the dynamics of these two types of granules. We performed time-lapse live-cell imaging on cells overexpressing CFP–MLN51, taking one picture every 0.5 seconds over 5 minutes (Fig. 6A; supplementary material Movie 1). We observed that, contrary to the MLN51-containing stress granules that did not move, SMIGs showed directed movements. Calculation of the mean squared displacement was performed on 119 SMIGs, and this showed that, when mobile, SMIGs show directed movements with an average speed of 3.7 µm/s over an average distance of 4 µm (Fig. 6B). It has been shown previously that microtubules are required for stress granule assembly (Ivanov et al., 2003). As SMIGs show directed movements, it would be tempting to hypothesize that these movements are involved in subsequent stress granule assembly. To test this hypothesis, we investigated the effects of microtubule disassembly on MLN51 localization and dynamics. At 24 hours after transfection, cells expressing the full-length CFP–MLN51 protein were treated for 2 hours with nocodazole, an antimitotic agent known to alter microtubule assembly. Although we found that, following treatment, the number of cells containing SMIGs decreased from 77.8% to 46.2% – similar to results that have been described previously for stress granules (Ivanov et al., 2003) – live-cell imaging failed to show any sort of SMIG fusion as a prerequisite for stress granule formation (data not shown). Taken together, our results reinforce the notion that the SMIGs are certainly distinct cytoplasmic entities rather than nascent stress granules.
We next wanted to see whether the P-body count was microtubule dependent. HeLa cells were transfected with CFP–MLN51. After 24 hours of expression, cells were treated for 2 hours with nocodazole and P-bodies were detected with anti-Dcp1a. As already established (Sweet et al., 2007; Aizer et al., 2008), nocodazole treatment increases the number of P-bodies by a factor of 3 – 8.23 P-bodies per cell following nocodazole treatment (n = 55) versus 2.66 P-bodies per cell (n = 53) in untransfected cells. It also strikingly reduces the effect of MLN51 overexpression on the number of P-bodies. Following nocodazole treatment, the average number of P-bodies per cell was reduced by a factor of 1.5 compared with the 3.5-fold reduction observed in untreated cells, and the percentage of cells having P-bodies is similar to that of untransfected cells. Interestingly, after nocodazole treatment, we observed a fourfold increase in contacts between SMIG and P-body markers (27% formed contacts in nocodazole-treated cells versus 7% in untreated cells). Taken together, these results suggest that overexpression of MLN51 induces the disassembly of P-bodies in a microtubule-dependent manner.
P-body disassembly is observed in HER2+ breast cancer cells that overexpress MLN51
In 50% of HER2+ breast cancers, MLN51 is co-amplified with c-Erbb2 (the gene encoding HER2) and the other genes of the amplicon (Bièche et al., 1996; Degot et al., 2002; Arriola et al., 2008). We analyzed the number of P-bodies in HER2+ breast cancer cells that either did or did not overexpress MLN51 [HER2+/MLN51+ (patient 204817) and HER2+/MLN51− (patient 20416E), respectively]. We compared this with the number of P-bodies in normal breast cells (patient P12/28 821) and HER2-negative cells [HER2−/MLN51−, a breast tumor with a HER2 immunohistochemistry (IHC) score of 0 (patient 207861)]. Overexpression of MLN51 was assessed by immunofluorescence (Fig. 7) and by IHC (supplementary material Fig. S4). P-bodies were counted following detection by immunofluorescence using anti-Dcp1a antibodies (Fig. 7). For HER2+/MLN51+ samples, we focused on cells that showed higher levels of overexpression of MLN51 (128 of 1380 cells; Fig. 7D; supplementary material Fig. S4D). Noticeably, in agreement with our observation in HeLa cells, the average number of P-bodies per cell was decreased in HER2+ breast cancer cells overexpressing MLN51 as compared with that of normal breast cells [2.08 P-bodies per cell in HER2+/MLN51+ cells (n = 128) versus 3.53 P-bodies per cell in normal cells (n = 1299)]. It was also decreased when compared with the number in HER2+ breast cancer cells that did not overexpress MLN51 [3.7 P-bodies per cell (n = 910)]. And finally, it decreased when compared with the number of P-bodies in HER2− cells [3.99 P-bodies per cell in HER2−/MLN− (n = 1103)]. The percentage of cells containing less than one P-body is strongly increased in HER2+/MLN51+ samples (43% compared with 18% in normal breast cells), whereas the percentage of cells with more than six P-bodies was decreased in the HER2+/MLN51+ sample versus the normal sample (2% compared with 19%). Finally, overexpression of MLN51 resulted in its localization into small cytoplasmic granules that resemble SMIGs (Fig. 7D, inset). Breast cancer cells that do not overexpress MLN51 show a similar distribution of P-body number to that found in normal breast cells (Fig. 7E). These results show that, in HER2+ breast cancer cells that overexpress MLN51, the number of P-bodies is reduced in a similar manner to that observed after overexpression of MLN51 in HeLa cells.
MLN51 is the only core EJC component that localizes mainly in the cytoplasm. In this study, we show for the first time that endogenous MLN51 is a component of P-bodies. It has been suggested that a small part of MLN51 is associated with polysomes (Degot et al., 2004). An interesting question is whether MLN51 localizes independently of the EJC or is only present in EJC-associated P-bodies. The absence of other core EJC components (eiF4A3, Magoh and Y14) from P-bodies (Chan et al., 2004; Chuang et al., 2013) favors the hypothesis of an independent mode of localization. Chuang et al. recently demonstrated that, in addition to its role in the EJC core, Y14 also has other important functions, and its overexpression leads to an increase in the number of P-bodies (Chuang et al., 2013). This Y14 action is opposite to that reported here for MLN51. Although we did not see that Y14 overexpression in HeLa cells had an effect on P-body numbers, both studies suggest that, following pioneer rounds of translation, several conformational changes might occur within the EJC core, thus allowing some of its factors to play additional roles in mRNA metabolism. However, it has to be noted that, as with Y14, the MLN51 region that is involved in non-EJC function is not the same as the one involved in its recruitment to the core complex ((Degot et al., 2004), this report). Moreover, we show that the effect of MLN51 overexpression on P-bodies can be reproduced when overexpressing a mutated MLN51 that is unable to bind to the EJC. This suggests that MLN51 can assemble on mRNAs independently of the EJC. Further studies will be necessary to elucidate the molecular mechanisms that allow them to perform these extra functions.
When overexpressed, MLN51 localizes in stress granules, but also mainly assembles in new and currently uncharacterized small cytoplasmic granules that we have named ‘small MLN51-induced granules’ (SMIGs). SMIGs contain poly(A)+ mRNA and show directed movements. SMIGs might be stress granule precursors, small P-bodies or an entirely new type of granule. Here, we show that SMIGs and stress granules differ in composition and kinetics. Using immunofluorescence, we could not detect a common P-body marker (such as Dcp1a, p54, Ge1 or Xrn1) in SMIGs. These observations favor the third hypothesis and suggest that these are a new type of RNA-containing granule. Other components of SMIGs are now under investigation. Interestingly, MLN51 is the human homolog of fly and mouse barentsz. Functional genetic screening has shown that barentsz is essential for oskar mRNA localization (van Eeden et al., 2001). Barentsz is also located in Staufen1 particles, which transport mRNAs along microtubules in the dendrites of mouse hippocampal neurons (Macchi et al., 2003). Dynamics studies have shown that to localize in a P-body a target must find its own way rather than relying on the active movement of a P-body to reach it (Aizer and Shav-Tal, 2008). It has also been reported that mRNA is required for cytoplasmic P-body maintenance (Cougot et al., 2004; Andrei et al., 2005; Ferraiuolo et al., 2005). Furthermore, a recent study showed that after nocodazole treatment, HIF-1α mRNA, which is overexpressed in solid tumors, accumulates in P-bodies. It then associates with Ago2, a member of the miRNA pathway, thus leading to translational repression (Carbonaro et al., 2011). Finally, it has been recently shown that in rat brains, barentsz associates with mRNAs, translational regulators and factors involved in miRNA-mediated translational repression in RNA granules (Fritzsche et al., 2013). Here, we show that SMIGs induced by MLN51 overexpression lead to the disassembly of mammalian P-bodies in a microtubule-dependent manner. Thus, one attractive model would be that, when overexpressed, MLN51 is able to take some repressed mRNAs out of P-bodies, and that this causes P-body disassembly. Besides its EJC function, therefore, MLN51 would also act as a transport factor for certain repressed mRNAs, being involved in their recycling out of the P-body via the microtubule network. In such a model, SMIGs would be RNA transport granules that induce P-body disassembly by titrating cytoplasmic RNA out of these structures in a microtubule-dependent manner. Nocodazole treatment would thus trap the SMIGs in P-bodies, preventing overexpression of the SMIG-associated transcripts. A more complete characterization of SMIGs will help us to better understand their function.
Interestingly, MLN51 is overexpressed in malignant breast cancer cells (Tomasetto et al., 1995; Degot et al., 2002; Arriola et al., 2008). MLN51 has been mapped to the q11–q21.3 region on the long arm of chromosome 17, a region also containing the c-Erbb2 gene (Tomasetto et al., 1995). The c-Erbb2 oncogene codes for the protein HER2, a tyrosine kinase belonging to the epidermal growth factor receptor family. Importantly, HER2 is overexpressed in ∼30% of human breast cancers, where it must play a major role in pathogenesis, because its overproduction correlates with tumor chemo-resistance and poor patient prognosis (Klapper et al., 2000; Degot et al., 2002). As part of the HER2 or Erbb2 amplicon, MLN51 is co-overexpressed in ∼50% of HER2+ breast cancer cells (Degot et al., 2002; Arriola et al., 2008). Similar to what is observed in HeLa cells after MLN51 overexpression, here, we show that P-bodies are also disassembled in MLN51-overexpressing HER2+ breast cancer cells. Maturation of an miRNA subset is impaired in tumor cells (Thomson et al., 2006), and impairing the miRNA processing pathway promotes tumorigenesis (Kumar et al., 2007). One attractive model would be that overexpression of MLN51 in HER2+ breast cancer cells leads to deregulation of the mechanism of miRNA-mediated translational repression. A remaining question is whether the general miRNA translation repression pathway is perturbed in cancer cells, or whether this effect is restricted to certain subclasses of miRNA-repressed mRNAs localized in P-bodies. The first step then will be to identify which mRNAs can be trapped in and released from P-bodies following overexpression of MLN51.
PML nuclear bodies are an example of a non-membranous organelle that has been found to show cancer-associated changes. They are thought to have a link to cancer because of the observation that the PML–RARα fusion protein disrupts nuclear bodies and contributes to leukemogenesis. Interestingly, treatment with arsenic or retinoic acid induces PML–RARα degradation but also leads to nuclear body reassembly, thus suggesting that the presence or absence of nuclear bodies correlates with cell malignancy (de Thé et al., 2012). In this paper, we discuss P-body disassembly following overexpression of MLN51, a newly identified component of P-bodies. It is tempting to speculate that, as is the case for nuclear bodies, P-body status could mirror cell malignancy. We are currently investigating P-body status in cancer cells at various stages of tumorigenesis.
MATERIALS AND METHODS
Cell culture and transfections
HeLa cells were maintained in DMEM GlutaMAX (Invitrogen, Paisley, UK) containing 10% fetal calf serum (FCS) and 1× penicillin-streptomycin (Life Technologies). For transient transfection in six-well plates, 100–200 ng of plasmid DNA were transfected using Lipofectamine Plus Reagent (Invitrogen) following the manufacturer's protocol. For treatment with nocodazole (Sigma-Aldrich, St Louis, MO), the cells were treated with 30 µM nocodazole for 2 hours at 24 hours after transfection. Cell sorting was performed on a FACSVantage SE option DiVa (BD Biosciences). Purity was >98%. Help for cell sorting was provided by C. Ebel (IGBMC).
To generate the anti-MLN51 rabbit polyclonal antibody, the synthetic peptide corresponding to residues 1–283 of the human MLN51 protein sequence was coupled to ovalbumin and injected into New Zealand rabbits. The rabbit anti-Dcp1a antibody was kindly supplied by Bertrand Séraphin (CNRS/IGBMC, Illkirch, France). The mouse anti-Dcp1a and anti-tubulin antibodies were from Abnova (Taipei City, Taiwan) and Sigma-Aldrich, respectively. Anti-MLN51 and -eIF4A3 antibodies were described previously (Degot et al., 2004; Daguenet et al., 2012). Anti-TIA1 and anti-Ge1 were from Santa Cruz Biotechnology (Dallas, TX) and anti-DDX6, anti-ATXN2 were from INTERCHIM (Montluçon, France). Cy3- and FITC-conjugated goat anti-mouse-IgG or anti-rabbit-IgG antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).
HeLa cells were seeded on glass coverslips and grown to 70% confluence. They were then washed with 1× PBS and fixed for 20 minutes at room temperature in 4% paraformaldehyde (PFA) in 1× PBS. After several washes with 1× PBS, cells were permeabilized with 0.1% Triton X-100 in 1× PBS for 10 minutes, then saturated for 30 minutes with 0.2% bovine serum albumin (BSA) in 1× PBS. After three washes with 1× PBS, cells were incubated for 1 hour at room temperature with the primary antibodies (1∶1000). After another three washes with 1× PBS, cells were incubated for 1 hour with the appropriate Cy3- or FITC-conjugated secondary antibodies (1∶1000). Cells were then washed and slides were mounted on Vectashield Medium with DAPI for nuclear counterstaining (Vector Laboratories, Burlingame, CA). Observations were made with an AxioImager Z1 upright microscope using Axiovision software (Zeiss, Oberkochen, Germany).
Fluorescence in situ hybridization
Poly(A)+ RNAs were detected by in situ hybridization (Bühler et al., 2002). Cells were grown on coverslips for 48 hours, fixed with 4% PFA for 30 minutes, then permeabilized overnight at 4°C with 70% ethanol. Cells were then rehydrated with 2× SSC containing 50% formamide, and hybridization was carried out with 300 ng Cy5-labeled oligo d(T) in 30% formamide in a humidified dark chamber at 37°C for 3 hours. After three washes with 2× SSC containing 30% formamide, the coverslips were subjected to standard immunofluorescence, mounted and analyzed as described above.
Immunofluorescence on HER2+ and normal breast cells
Paraffin-embedded tissue was cut into 4-µm-thick sections, mounted on positively charged slides, then dried at 58°C for 60 minutes. Immunohistochemical staining was performed on the Discovery XT Automated IHC staining system (Ventana, Tucson, AZ). Following deparaffination with Ventana EZ Prep solution at 75°C for 8 minutes, antigen retrieval was performed at 95–100°C for 48 minutes using CC1, a proprietary Ventana Tris-based buffer solution. After rinsing, slides were incubated at 37°C for 60 minutes with a 1∶50 dilution of rabbit anti-MLN51 and 1∶50 mouse anti-Dcp1a. Detection was performed using Alexa-Fluor-555-conjugated goat anti-rabbit-IgG and the Alexa-Fluor-488-conjugated goat anti-mouse-IgG (both from Invitrogen) for 45 minutes at room temperature. DAPI staining was performed in order to visualize the nuclei. Images were acquired on a Nikon Ni-E microscope and deconvoluted using the NIS-Elements AR software.
Paraffin-embedded tissue was cut into 4-µm-thick sections, mounted on positively charged slides, then dried at 58°C for 60 minutes. Immunohistochemical staining was performed on the Discovery XT Automated IHC staining system using the Ventana DAB Map detection kit. Following deparaffination with Ventana EZ Prep solution at 75°C for 8 minutes, antigen retrieval was performed at 95–100°C for 48 minutes using Ventana CC1. Endogenous peroxidase was blocked with Inhibitor-D 3% H2O2 (Ventana) for 4 minutes at 37°C. For MLN51 staining, the slides were incubated after rinsing at 37°C for 60 minutes with a 1∶100 dilution of rabbit anti-MLN51. Signal enhancement was performed using the Ventana DAB Map Kit. Biotinylated goat anti-rabbit-IgG (H+L) was used as the secondary antibody (Vector). For HER2 detection, slides were incubated at 37°C for 20 minutes with a 1∶1000 dilution of polyclonal rabbit antibody against human HER2 (Dako, Glastrup, Denmark). Signal enhancement was performed using the UltraView Universal DAB (Ventana) and goat anti-rabbit-IgG HRP. Slides were then counterstained for 8 minutes and manually dehydrated before coverslips were added. Slides were treated with a Hamamatsu NanoZoomer 2.0-RS scanner and images were visualized with the NDP.view 1.2.47 software.
peCFP-MLN51 (Degot et al., 2002; Degot et al., 2004) were transfected into HeLa cells. At 24 hours post-transfection, acquisition was performed on a Nikon TE200 widefield microscope (100× NA 1.45 objective) equipped with an EM-CCD camera (Cascade 512B, Roper Scientific). Z-stacks were captured every 0.5 s for 5 minutes. The Z-stacks were used to assemble movies using MetaMorph software (Molecular Devices, Sunnyvale, CA).
High-pressure freezing and freeze substitution
A cell slurry was made by resuspending HeLa cells in 20% BSA in DMEM. A 2-µl drop of the sample was loaded onto a pre-heated, 0.5-mm-thick flat specimen carrier. This had gold-plated surfaces and a cavity diameter of 1.5 mm and depth of 0.2 mm (catalog number 16706898, Leica Microsystems AG, Wetzlar, Germany). Cells were frozen in a Leica EM PACT2 high-pressure freezer. Samples were transferred to an automatic freeze-substitution system (Leica EM AFS2) equipped with an automatic reagent handling system (EM FSP), then placed in the precooled −90°C substitution solution. Cells were freeze-substituted in acetone containing 0.1% uranyl acetate at −90°C for 60 hours. The temperature was then increased to −50°C at a rate of 3°C per hour, and samples were kept at this temperature for 24 hours. The samples were then washed once with acetone and three times with 100% ethanol at −50°C. Samples were infiltrated with resin/100% ethanol mixtures by raising the volume-to-volume proportion of Lowicryl HM20 as follows: 2 hours at 25%, 2 hours at 50%, overnight at 75% and, finally, 1 hour (four times) at 100%. Finally, polymerization was carried out at −50°C for 48 hours and at 20°C for 48 hours. Ultrathin 90–110 nm sections were collected on 300-mesh nickel grids.
Lowicryl HM20-embedded thin sections were first blocked with 1% serum albumin in 100 mM pH 7.4 phosphate buffer for 2.5 hours, then allowed to react for 2 hours with primary antibody diluted in blocking buffer. The dilutions were 1∶250 for rabbit and mouse anti-Dcp1a, and for rabbit anti-MLN51. Following four 10-minute washes with blocking buffer, and one wash in 100 mM Tris-HCl pH 7.4 buffer containing 1% serum albumin, grids were incubated for 1 hour with the goat anti-rabbit-IgG or anti-mouse IgG antibody coupled to gold particles at a dilution of 1∶40 (Delta Microscopies, Ayguesvives, France). After subsequent washes with the same buffer, grids were fixed with 2.5% glutaraldehyde and contrasted for 1.5 hours in 5% uranyl acetate. Double immuno-detections consisted of repeating two single immuno-detections in a row: a first detection using rabbit anti-MLN51 revealed with goat anti-rabbit IgG conjugated to 6-nm gold particles, and a second detection using mouse anti-Dcp1a revealed with goat anti-mouse IgG conjugated to 15-nm gold particles.
Human patient samples
Human patient samples were analyzed as described above. The study cohort was made up of ten patients and included two samples of non-malignant breast cells, three HER2− and five HER2+ breast cancer samples. This study was approved by the institutional boards of the Centre de Ressources Biologiques Santé de Rennes.
We are grateful to Bruno Turlin, Florence Godey and Mireille Dugast (Centre de Ressources Biologiques Santé de Rennes, France) and Jean Levêque (Département de Gynécologie et d'Obstétrique, Centre Hospitalier Universitaire de Rennes, France) for invaluable help with the clinical part of the present investigation. We would like to thank Bertrand Séraphin (Institut Génétique Biologie Moléculaire Cellulaire, Strasbourg, France) for providing us with the anti-Dcp1a antibody and to Juliana Berland for insightful comments on the manuscript. Electron microscopy experiments were processed at the Microscopy Rennes Imaging Center (MRic) facilities (http://microscopie.univ-rennes1.fr/).
N.C. conceived and performed the experiments; analyzed data and wrote the manuscript. E.D. performed experiments with immunofluorescence; A.B. performed experiments with immunofluorescence, before and during the revision process; A.C. performed experiments with high pressure freezing and freeze substitution; D.T. performed experiments with immunoelectron microscopy; P.B. and A.F. performed experiments with human patient samples; F.G. designed and selected the study cohort; E.B. conceived experiments and performed videomicroscopy data analysis; C.T. conceived experiments, performed immunofluorescence and W.B. experiments during the revision process; R.G. directed the project and analyzed data. N.C. wrote the manuscript with advice and comments from the other authors. All the authors contributed to the design and interpretation of experiments.
This work was supported by the French National Agency [grant numbers ANR-08JCJC-0027-01 and ANR-09-MIE]; the Institut Universitaire de France; and the Ligue Nationale Contre le Cancer. E.D. had a fellowship from the Fondation pour la Recherche Médicale (FRM). Funds to C.T. were from The Ligue Nationale Contre Le Cancer (CCIR-GE).
The authors declare no competing interests.