ABSTRACT
Progesterone mediates its physiological functions through activation of both transcription-coupled nuclear receptors and seven-pass-transmembrane progesterone receptors (mPRs), which transduce the rapid non-genomic actions of progesterone by coupling to various signaling modules. However, the immediate mechanisms of action downstream of mPRs remain in question. Herein, we use an untargeted quantitative proteomics approach to identify mPR interactors to better define progesterone non-genomic signaling. Surprisingly, we identify the very-low-density lipoprotein receptor (VLDLR) as an mPRβ (PAQR8) partner that is required for mPRβ plasma membrane localization. Knocking down VLDLR abolishes non-genomic progesterone signaling, which is rescued by overexpressing VLDLR. Mechanistically, we show that VLDLR is required for mPR trafficking from the endoplasmic reticulum to the Golgi. Taken together, our data define a novel function for the VLDLR as a trafficking chaperone required for the mPR subcellular localization and, as such, non-genomic progesterone-dependent signaling.
This article has an associated First Person interview with the first author of the paper.
INTRODUCTION
Progesterone (P4) is a steroid hormone that regulates various reproductive processes in females, including ovulation, implantation and sexual differentiation. Signaling downstream of P4 has been studied primarily through the activation of nuclear receptors that act as transcription factors to stimulate P4-dependent gene expression (Ellmann et al., 2009). However, P4 is also known to transduce rapid non-genomic signals that link to cAMP, Ca2+ and the mitogen-activated protein kinase (MAPK) cascade, among other signaling modules (Valadez-Cosmes et al., 2016). Initially, activation of these signaling cascades was attributed to the classical ability of the progesterone receptor to interact with cytoplasmic factors (Dressing et al., 2011). However, non-genomic actions of progestins have been demonstrated in several tissues that do not express classical progesterone receptors, and P4 coupled to bovine serum albumin (BSA), which is unable to cross the plasma membrane and interact with the classical P4 receptor, is nonetheless effective at mediating non-genomic P4 signaling (Bandyopadhyay et al., 1998; Dressing et al., 2011; Peluso et al., 2002). These results argued for the presence of membrane P4 receptors that are distinct from the nuclear P4 receptors. In 2003, the Thomas laboratory identified a family of membrane progesterone receptors (mPRs) from fish ovaries (Zhu et al., 2003a,b) that belong to the progestin and adiponectin (AdipoQ) receptor family (also named PAQ receptors). However, the signal transduction cascade downstream of mPRs that mediates the non-genomic actions of P4 remains unclear.
The non-genomic action of mPR and the ensuing signaling cascade have been studied extensively over the years in frog and fish oocytes, where progesterone releases oocyte meiotic arrest. Vertebrate oocytes must undergo a maturation period before they become fertilization competent and are able to support embryonic development. Oocyte maturation encompasses progression through meiosis and arrest at metaphase II until fertilization (Machaca, 2007). It is defined by drastic cellular remodeling characterized by the dissolution of the nuclear envelope (referred to as germinal vesicle breakdown; GVBD), extrusion of the first polar body, chromosome condensation and arrest in metaphase of meiosis II (Bement and Capco, 1990; Nader et al., 2013; Sadler and Maller, 1985; Smith, 1989; Voronina and Wessel, 2003). Prior to oocyte maturation and for prolonged periods of time, fully grown vertebrate oocytes are arrested at prophase of meiosis I, as they grow and stock molecular components essential for future development (Smith, 1989; Voronina and Wessel, 2003). P4 releases Xenopus oocyte meiotic arrest by ultimately activating maturation-promoting factor (MPF; the cyclin-B–cdc2 complex), the key driver of meiosis (Nader et al., 2013; Nebreda and Ferby, 2000). In amphibian oocytes, treatment with the membrane-impermeant P4 conjugated to BSA (P4–BSA), efficiently releases meiotic arrest. These and other results indicate that P4 acts through a membrane receptor rather than the classical nuclear receptor (Bandyopadhyay et al., 1998; Blondeau and Baulieu, 1984; Josefsberg Ben-Yehoshua et al., 2007), although there is evidence for a potential role for the nuclear receptor as well (Bayaa et al., 2000; Tian et al., 2000). There is also evidence that other steroids release frog oocyte meiotic arrest, with testosterone being the physiological inducer of oocyte maturation in vivo (Lutz et al., 2001). In 2007, Ben-Yehoshua et al. reported the identification of a Xenopus laevis mPR ortholog, mPRβ (also known as PAQR8, herein referred to as mPR for simplicity) as the receptor functionally responsible for the resumption of meiosis in response to P4 (Josefsberg Ben-Yehoshua et al., 2007). It is well documented that high levels of cAMP and protein kinase A (PKA) are important in maintaining meiotic arrest in vertebrates (Bravo et al., 1978; Cho et al., 1974; Conti et al., 2002; Gallo et al., 1995; Lutz et al., 2000; Maller and Krebs, 1977; Meijer and Zarutskie, 1987; Sadler et al., 2008; Sheng et al., 2001; Stern and Wassarman, 1974). In Xenopus oocytes, the high levels of cAMP–PKA are maintained, at least in part, through the action of the constitutively active Gαs-coupled G protein-coupled receptor GPR185 (Ríos-Cardona et al., 2008), the homolog of mammalian GPR3, which has been found to block oocyte maturation in mouse oocytes (Freudzon et al., 2005; Mehlmann et al., 2002, 2004; Norris et al., 2007). mPR activation does not seem to signal through Gαi to inhibit adenylate cyclase since treatment with pertussis toxin was ineffective at blocking P4-induced oocyte maturation (Mulner et al., 1985; Olate et al., 1984; Sadler et al., 1984). Similarly, maturation in mouse oocytes also seems to be independent from the action of the Gi subunit (Mehlmann et al., 2006). Moreover, although high levels of cAMP are essential in maintaining oocyte meiotic arrest, oocyte maturation progresses without any changes in cAMP or PKA levels (Nader et al., 2016). These findings argue for a positive signal downstream of mPR that overrides the cAMP–PKA inhibitory signal (Nader et al., 2016). This positive signal is likely to be initiated through the P4–mPR axis to release the oocyte from meiotic arrest.
Here, we employ an untargeted proteomics approach to identify the mPR interactome to better define signaling downstream of mPR. We identify the very-low-density lipoprotein receptor (VLDLR) as an mPR-interacting protein, and show that VLDLR is essential for mPR plasma membrane localization, and, as such, its signaling function. VLDLR acts as a molecular chaperone that is required for the trafficking of mPR from the endoplasmic reticulum (ER) to the Golgi. In the absence of VLDLR, mPR concentrates in the ER and does not reach the cell membrane where it performs its signaling function. Hence, the VLDLR plays an essential role in regulating mPR trafficking and signaling during oocyte maturation.
RESULTS
GFP-tagged mPR – functional analysis and localization
To characterize the subcellular localization of the Xenopus mPRβ (mPR) in the absence of good anti-mPR antibodies, we tagged mPR with GFP at its N- (GFP–mPR) or C-terminus (mPR–GFP). In order to test the functionality of the GFP-tagged mPR, the physiological effects of overexpressed untagged mPR, GFP–mPR and mPR–GFP were tested on progesterone (P4)-induced oocyte maturation as compared to the expression of GFP alone (Fig. 1). In accordance with a previous report (Josefsberg Ben-Yehoshua et al., 2007), we found that overexpressing wild-type mPR significantly (P=0.045) potentiated oocyte maturation at suboptimal concentrations of P4 concentration (3×10−8 M) (Fig. 1A). Similarly, the C-terminally tagged mPR, mPR–GFP was able to potentiate oocyte maturation at suboptimal P4 concentrations (Fig. 1A, P=0.035). In contrast, no potentiation of oocyte maturation was observed when GFP alone or the N-terminally tagged GFP–mPR proteins were overexpressed (Fig. 1A). Although the maximum GVBD percentage was not significantly affected by expression of the different mPR constructs when an optimal concentration of 3×10−7 M P4 was used (Fig. 1B), oocytes overexpressing mPR and mPR–GFP consistently reached 50% GVBD around 3 h faster (P=0.048 and P=0.022, respectively) when compared to oocytes overexpressing GFP alone or GFP–mPR (Fig. 1C,D). Both GFP–mPR and mPR–GFP were expressed at similar levels (Figs 1E and 2B). These results show that the C-terminally tagged mPR is functional and replicates the activity of the wild-type protein. In contrast, tagging mPR at its N-terminus somehow interferes with its function, resulting in a functionally defective protein.
We then checked the trafficking of the two constructs GFP–mPR and mPR–GFP by performing confocal imaging of live oocytes. To mark the plasma membrane, we also expressed the resident plasma membrane (PM) protein TMEM–mCherry, which encodes a Ca2+-activated Cl− channel (Yu et al., 2010). A z-stack of confocal images was acquired across the oocyte (Fig. S1), with the orthogonal section through this stack representing the subcellular distribution of the proteins as previously described (Yu et al., 2010). We have previously shown that TMEM–mCherry targets to the PM and that its membrane residence is not affected throughout oocyte maturation (Nader et al., 2014; Yu et al., 2010). Interestingly, the GFP–mPR construct, which is unable to release meiotic arrest, was mostly intracellular and unable to localize to the PM (Fig. 1E; Fig. S1B), whereas a significant proportion of the functional mPR–GFP localized to the PM (microvilli) in addition to localizing intracellularly (Fig. 1E; Fig. S1A).
GFP–mPR- and mPR–GFP-expressing oocytes were stained with wheat germ agglutinin (WGA) to label the plasma membrane, allowing for the quantification of the distribution of mPR at the PM (Courjaret et al., 2016; El-Jouni et al., 2008). mPR–GFP was found to be significantly enriched at the PM (38.91±1.89%, mean±s.e.m.; n=10), whereas GFP–mPR was mainly restricted to the cytosol, with only 16.18±0.98% (n=12) at the PM (Fig. 1F). The intracellular component of mPR–GFP could represent the proportion of the receptor that is undergoing biogenesis, and/or saturation of the trafficking machinery when mPR is overexpressed. Furthermore, our conservative quantification approach using the peak of WGA staining as defining the PM (see Materials and Methods), is likely to underestimate the intracellular proportion of both GFP-tagged mPRs given the opacity and size of the oocyte. Nevertheless, the subcellular localization data clearly show that GFP–mPR is defective in its trafficking to the PM, which correlates with its inability to accelerate oocyte maturation when overexpressed (Fig. 1A–D). This argues that mPR exerts its signaling function only when it localizes to the PM and is unable to signal when localizing intracellularly. Furthermore, it shows that mPR does not signal constitutively when overexpressed and still requires P4.
Identification of VLDLR as an mPR-interacting protein
The GFP–mPR and mPR–GFP constructs provide perfect tools to define mPR-interacting proteins that mediate its non-genomic actions. Building on the fact that GFP–mPR is not functional, whereas mPR–GFP mediates oocyte maturation with a similar efficiency to that of the untagged receptor, we used a quantitative unbiased proteomics approach to identify proteins that preferentially interact with the functional mPR–GFP (Fig. 2A). Proteins that interact preferentially with mPR–GFP as compared to the non-functional GFP–mPR are likely to be important for its trafficking and/or signaling functions. Proteomics approaches are notorious for identifying potential non-specific proteins given the biochemical approaches used and the dynamic range of the mass spectrometry (MS) approach itself. To minimize these potential artifacts, we pulled down GFP–mPR or mPR–GFP from oocytes overexpressing either construct using anti-GFP-linked beads (Fig. 2A). This avoids contamination with the endogenous mPR and isolates the interactome of the specific overexpressed protein. Both constructs were expressed to similar levels and were pulled-down efficiently (Fig. 2B), followed by trypsin digestion and dimethyl labeled with heavy (mPR–GFP) or light (GFP–mPR) isotopes before MS analysis as outlined in Fig. 2A. The ‘heavy-over-light’ ratio identified relatively few candidates that preferentially bind the functional mPR–GFP as compared to GFP–mPR (Fig. 2C; Table S1). VLDLR emerged as the most consistently and highly enriched protein (Fig. 2C; Table S1). We therefore focused on the VLDLR as a potential mediator of P4-dependent signaling through mPR to drive the oocyte maturation and meiosis progression.
To further characterize the role of VLDLR, we engineered N- and C-terminally mCherry-tagged VLDLR constructs. In order to validate the quantitative proteomics results and confirm the association of VLDLR with the C-terminally tagged mPR–GFP, we co-expressed VLDLR tagged with mCherry at either the N- (Ch–VLDLR) or the C-terminus (VLDLR–Ch) with mPR–GFP and performed an immunoprecipitation using anti-GFP beads (Fig. 2D). Immunoprecipitation of mPR–GFP pulled down VLDLR–Ch and Ch–VDLDR, indicating that the two proteins are part of the same complex in situ (Fig. 2D).
VLDLR is part of the low-density-lipoprotein (LDL) transmembrane receptor family that localizes to the PM (Go and Mani, 2012). Consistent with this, the Xenopus VLDLR tagged at either end localized to the PM, as indicated by the WGA co-staining (Fig. 2E). Although both VLDLR constructs primarily localize to the PM (Fig. 2E), we observed a modest but significant (P=0.012) enrichment of the N-terminally tagged Ch–VLDLR (67.31± 1.876%; mean±s.e.m.; n=29) at the PM as compared to VLDLR–Ch (58.47±2.837%; n=29) (Fig. 2F).
Role of VLDLR in P4–mPR-induced maturation
To directly test the role of VLDLR in P4–mPR signaling, we knocked down VLDLR expression and tested the effects on oocyte maturation. Antisense oligonucleotides were effective at knocking down VLDLR RNA levels as compared to the control sense oligonucleotides (Fig. 3A). Although the sense oligonucleotides resulted in some decrease in VLDLR RNA levels as compared to that found in control uninjected oocytes, the antisense oligonucleotides almost completely abrogated VLDLR RNA (Fig. 3A). The antisense effect was specific to VLDLR as no significant changes in the levels of mPR RNA were detected (Fig. 3A). The antisense oligonucleotides were also effective at blocking exogenous VLDLR–Ch protein expression (Fig. 3B).
Interestingly, knocking down VLDLR expression strongly and significantly (P<0.0001) inhibited P4-induced oocyte maturation (Fig. 3C). The inability of oocytes to mature following VLDLR knockdown was confirmed biochemically by assaying the activation of both MAPK (ERK1/2) and MPF, two kinases known to be activated downstream of P4 treatment and are required for oocyte maturation (Fig. 3D). The MAPK cascade is activated downstream of progesterone and contributes to the induction of MPF (cyclin B–Cdc2) the master regulator of oocyte maturation (Castro et al., 2001; Palmer and Nebreda, 2000). MAPK activation is detected by phosphorylation, whereas MPF activation is followed by dephosphorylation of the kinase subunit Cdc2 (Fig. 3D). Tubulin was used as a loading control.
To confirm that the antisense effect is specific to the knockdown of the VLDLR and not due to some offsite effects, we tested whether overexpression of the VLDLR can rescue the VLDLR knockdown. Overexpressing untagged VLDLR (10 ng/oocyte) completely reversed the inhibition of oocyte maturation mediated by VLDLR knockdown (Fig. 3E), confirming the specificity of VLDLR knockdown on P4-mediated maturation. In addition, untagged mPR overexpression (10 ng/oocyte) was also able to significantly rescue the inhibition of oocyte maturation mediated by VLDLR knockdown, from 17% of oocytes showing GVBD to 57% (Fig. 3E). Interestingly, the rescue of oocyte maturation was coupled to a significant increase of mPR PM levels, as discussed below (Fig. 4F). These data collectively show that VLDLR is required for oocyte maturation and to mediate the activation of the signaling cascade downstream of P4-mPR.
VLDLR is essential for mPR PM localization
The high enrichment of VLDLR within the functional mPR–GFP complex that traffics normally to the plasma membrane, as compared to the defective GFP–mPR that localizes intracellularly, hints to a role for VLDLR in regulating mPR trafficking. This would be functionally critical since the GFP–mPR, which is unable to reach the PM does not support oocyte maturation. To test whether VLDLR modulates mPR trafficking, we used a progesterone analog coupled to BSA and fluorescein (that is membrane impermeant because of the BSA moiety) that allows easy imaging and quantification of endogenous mPR at the PM. To validate P4–BSA–fluorescein as a reliable reagent to quantify mPR at the PM, we overexpressed mPR, which resulted in a significant (P=0.0042) increase in P4–BSA–fluorescein binding (Fig. 4A,B). In contrast, knocking down VLDLR expression (antisense) results in a significant (P<0.0001) 60% decrease of endogenous mPR localizing to the plasma membrane as compared to the sense control or naïve untreated oocytes (Fig. 4A,B). These data show that VLDLR is required for the trafficking of endogenous mPR to the PM.
To better define the mPR trafficking defect in the absence of VLDLR, we tested the effect of VLDLR knockdown on trafficking of overexpressed mPR. Oocytes were co-injected with RNAs that express mPR–GFP and TMEM–mCherry to mark the PM along with VLDLR sense or antisense oligonucleotides (Fig. 4C–E). The co-injection of mPR-encoding RNA with antisense oligonucleotides targeting VLDLR was designed to knockdown VLDLR expression at the same time as mPR was overexpressed over a 48 h timecourse to address the role of the VLDLR in the biogenesis and PM targeting of mPR. We found that VLDLR knockdown inhibits the trafficking of mPR–GFP to the PM and its colocalization with TMEM–mCherry as observed in the orthogonal confocal sections (Fig. 4C). This was also apparent from the shift of mPR–GFP fluorescence towards the interior of the cell following VLDLR knockdown (Fig. 4D). Quantification of overexpressed mPR PM residence in oocytes where VLDLR has been knocked down compared to the amount in the sense control, reveals a significant (P<0.0001) decrease, by ∼20%, in the amount of mPR at the PM (Fig. 4E). The fact that VLDLR knockdown results in a more substantial decrease (60%) in endogenous mPR at the PM than what is seen when mPR is overexpressed (20% decrease) argues for a saturation effect of the trafficking machinery with excess mPR, thus allowing mPR to reach the PM even when VLDLR is limiting. This is consistent with the rescue of oocyte maturation observed in antisense VLDLR-treated oocytes (Fig. 3D). To directly test this conclusion, we overexpressed mPR in oocytes where VLDLR has been knocked down (Fig. 4F). This resulted in a partial rescue of the ability of mPR to localize to the PM. Hence, excess mPR saturates the VLDLR-dependent trafficking regulation.
Although our data show a clear role for the VLDLR in regulating the trafficking of mPR, it is not clear whether this effect is due to a broad deregulation of the trafficking machinery or whether it is specific to mPR. The fact that VLDLR interacts with mPR would argue for a specific effect. However, to directly test whether VLDLR is a specific chaperone of mPR, we measured the effect of VLDLR knockdown on total PM surface area by determining the membrane capacitance. Knockdown of VLDLR had no significant effect on membrane capacitance, which measured 248+11.3, 243+ 9.7 and 231+8.2 nF (n=12 oocytes/condition, mean±s.e.m.) in naïve, VLDLR sense and VLDLR antisense-injected oocytes respectively. We further tested whether VLDLR affects the trafficking of Orai1, a PM Ca2+ channel that is required for store-operated Ca2+ entry (SOCE) and that we have previously shown recycles at the PM (Yu et al., 2010). Again, VLDLR knockdown had no significant effect on the SOCE current, arguing that it does not affect Orai1 trafficking (Fig. 4G). These results rule out a general effect of VLDLR on plasma membrane recycling and support a specific role for the VLDLR in regulating mPR trafficking to the PM.
Colocalization of mPR and VLDLR
VLDLR is a type 1 single-pass transmembrane protein with an extracellular N-terminus and cytosolic C-terminus. To further characterize the mPR–VLDLR interaction, we co-expressed and imaged the different combinations of N- and C-terminally tagged mPR and VLDLR (Fig. 5). Co-expression of the functional C-terminally tagged mPR–GFP with VLDLR tagged at either end, shows colocalization of both proteins at the PM (Fig. 5A,B). Furthermore, mPR–GFP colocalizes intracellularly with VLDLR–Ch, which shows a partial intracellular distribution (Fig. 5B). Ch–VLDLR is almost exclusively at the PM (Fig. 5A), whereas VLDLR–Ch can be clearly found intracellularly (Fig. 5B). By contrast, GFP–mPR does not colocalize with either VLDLR construct and exhibits a reticular intracellular distribution reminiscent of the ER (Fig. 5C,D). These results show that when mPR is tagged with GFP at its N-terminus it is unable to interact with VLDLR. This was confirmed by immunoprecipitation; pulling down GFP–mPR did not co-immunoprecipitate VLDLR (Fig. 5E). Remarkably, overexpressing VLDLR–Ch, which localizes intracellularly to a higher degree than Ch–VLDLR (Fig. 2E,F), slightly but significantly lowered the amount of endogenous and overexpressed mPR at the plasma membrane (Fig. 5F,G). In contrast, overexpressing Ch–VLDLR significantly increased the proportion of endogenous mPR at the plasma membrane, whereas no effect was found on overexpressed mPR, probably due to the saturation of the exocytosis machinery (Fig. 5F,G). These results show that when the VLDLR is able to interact with mPR this alters its membrane residence and trafficking. Collectively, our data validate VLDLR as a novel trafficking chaperone of mPR, and importantly, in that capacity, as a critical physiological regulator of oocyte maturation.
VLDLR is required for mPR trafficking from the ER to the Golgi
As shown in Fig. 5A,B, the intracellular distribution of mPR–GFP exhibits the typical reticular ER distribution (stars), in addition to a portion of the protein distributing to punctate structures that are reminiscent of the Golgi complex in the oocyte (arrowheads). In contrast to the tight reticular distribution typical of the ER in the frog oocyte (Figs 5 and 6), expression of GFP or mCherry alone shows a diffuse cytoplasmic distribution with exclusion from the areas containing pigment granules (Fig. S2A). Although, as far as we are aware, the Golgi has not been visualized in Xenopus oocytes using fluorescence probes, it was studied by electron microscopy in the 1980s. The oocyte, given its large size, has multiple Golgi complexes that are distributed both deep in the cytosol and within the cortex (Colman et al., 1985). As is the case in mammalian cells, the Golgi stacks fragment during M-phase in the oocyte (Colman et al., 1985). To determine whether the mPR-positive structures are indeed Golgi complexes, we expressed the Golgi-resident enzyme N-acetylgalactosaminyltransferase-2 (GalNac) fused to GFP (Storrie et al., 1998). To differentiate the Golgi complexes from the ER we co-expressed GalNac–GFP (Golgi) with an ER marker (KDEL–mCherry) (Fig. 6A). KDEL–Cherry shows the typical reticular ER distribution in the oocyte, whereas the Golgi marker GalNac–GFP distributes in distinct punctate structures (Fig. 6A, arrowheads) that do not overlap with the ER (Fig. 6A, top row). Furthermore, and consistent with the literature (Colman et al., 1985), the Golgi complexes fragmented during meiosis as illustrated by the dramatic decrease in the GalNac-positive structure in the mature egg (Fig. S2B). Co-expression of VLDLR with the Golgi marker GalNac–GFP shows that VLDLR–Ch localized to the Golgi (Fig. 6A, middle row, arrowheads). Therefore, VLDLR–Ch, in addition to its PM distribution (Fig. 5B), localizes to the Golgi intracellularly. The co-expression of mPR–GFP with VLDLR–Ch shows that mPR–GFP also localizes to the intracellular punctate structures (Fig. 6A, bottom row, arrowheads) identified in the GalNac–GFP/VLDLR–Ch co-expression experiment as Golgi (Fig. 6A, middle row). These results show that mPR–GFP colocalizes to the Golgi with VLDLR–Ch during its biogenesis.
Interestingly, the N-terminally tagged GFP–mPR does not colocalize with VLDLR–Ch at the Golgi and is instead enriched in the ER (Fig. 5C,D). GFP–mPR does not traffic to the plasma membrane (Fig. 5C), and does not interact with VLDLR in co-immunoprecipitation experiments (Fig. 5E). Furthermore, GFP–mPR is not functional in terms of signaling downstream of P4 in terms of releasing meiotic arrest (Fig. 1A), consistent with its inability to traffic efficiently to the PM (Fig. 1E). These results argue that GFP–mPR, given its inability to interact with VLDLR, is unable to traffic from the ER to the Golgi in transit to the PM. To directly test whether this is the case, we knocked down endogenous VLDLR expression using antisense oligonucleotides and quantified the localization of mPR to the Golgi puncta. If VLDLR is indeed required for mPR trafficking from the ER to the Golgi, we would expect a decreased Golgi localization of expressed mPR–GFP when endogenous VLDLR is knocked down. This is indeed what we observe. Injecting oocytes with the VLDLR sense oligonucleotides gave the typical punctate mPR-positive Golgi distribution (Fig. 6B, top row). In contrast, injecting oocytes with VLDLR antisense oligonucleotides resulted in a significant reduction of the localization of mPR to the Golgi puncta and a more pronounced ER localization (Fig. 6B, bottom row). Quantification of the number of mPR–GFP-positive Golgi puncta in the sense versus antisense VLDLR oligonucleotide-treated oocytes shows a significant reduction in the antisense VLDLR oocytes (7.97±1.06 Golgi puncta/100 µm2) as compared to their sense-injected counterparts (13.25±1.22 Golgi puncta/100 µm2) (Fig. 6C). To rule out the possibility that VLDLR knockdown has any effect on the Golgi directly, we quantified the number of Golgi puncta (GalNac–GFP positive) in sense and antisense VLDLR-injected oocytes, and show no difference in the number of Golgi puncta (Fig. 6D). This shows that the VLDLR is not required for Golgi stability but rather for the trafficking of mPR from the ER to the Golgi. Therefore, the decreased PM residence of mPR in the absence of the VLDLR is due to a defect in the transit of mPR from the ER to the Golgi.
DISCUSSION
The fast non-genomic progesterone signaling via membrane progesterone receptors has emerged as an important regulator of biological function in the nervous system, female and male reproductive tissues, and immune and cancer cells (Dosiou et al., 2008; Dressing et al., 2011; Moussatche and Lyons, 2012; Valadez-Cosmes et al., 2016). The first mPR was cloned from fish oocytes, followed by investigation for its role in oocyte maturation in multiple non-mammalian species (Josefsberg Ben-Yehoshua et al., 2007; Thomas et al., 2002; Zhu et al., 2003b). Since then, and given the broad expression profile for mPRs in different tissues in vertebrates, studies have elucidated important signaling roles for this gene family in diverse physiological and pathological processes (Dressing et al., 2011), including sperm motility (Thomas et al., 2009; Valadez-Cosmes et al., 2016), female reproduction (Qiu et al., 2008; Nutu et al., 2007, 2009) and immune cell function (Dressing et al., 2011).
The VLDLR is a transmembrane lipoprotein receptor of the low-density-lipoprotein receptor (LDLR) family, which consists of seven structurally closely related proteins (May et al., 2005). The LDLR family is composed of constitutively recycling cell surface receptors that are responsible for the uptake of lipoproteins and other ligands via endocytosis and lysosomal delivery, primarily in the liver (Gotthardt et al., 2000; May et al., 2005). However, some members of the family, particularly the VLDLR, also mediate intracellular signaling cascades especially in the brain (May et al., 2005; Ranaivoson et al., 2016). The VLDLR amino acid sequence is highly conserved during evolution with ∼95% identity between mammals, and more than 75–80% identity among vertebrates (humans, chickens and frogs), implying an essential conserved function in these species (Bujo and Yamamoto, 1996). In mammals, the VLDLR has a broad tissue distribution with a predominant expression in the central nervous system (Bujo and Yamamoto, 1996; May et al., 2005), suggesting that its primary function is separate from lipoprotein metabolism. In fact, VLDLR-knockout mice do not exhibit any defect in lipid homeostasis (Frykman et al., 1995), but rather exhibit neurodevelopmental defects reminiscent of a Reelin/Disabled-like disruption of neuronal migration (Trommsdorff et al., 1999). Consistent with this, the VLDLR acts as a canonical receptor for Reelin, and mediates its signaling functions (D'Arcangelo et al., 1999; Hiesberger et al., 1999; Ranaivoson et al., 2016; Trommsdorff et al., 1999). Naturally occurring mutations in humans strongly corroborate the role of VLDLR in the brain (Ali et al., 2012; Boycott et al., 2005; Moheb et al., 2008; Ozcelik et al., 2008; Türkmen et al., 2008). Unlike mammals, the chicken VLDLR expression is restricted to the oocyte (Bujo et al., 1994; Bujo and Yamamoto, 1996), and functions primarily in VLDL and vitellogenin uptake to support yolk accumulation and oocyte growth (Bujo et al., 1994). Consistent with this, mutations in the chicken VLDLR cause female sterility and severe hyperlipidemia (Bujo et al., 1995; Stifani et al., 1990).
In Xenopus, the VLDLR is involved in yolk accumulation in oocytes. However, based on its broad expression pattern (Okabayashi et al., 1996), it is likely to mediate other signaling functions in the frog, similar to what is observed in mammals. Opresko and Wiley showed a single class of low affinity (1.3×10−6 M) and high specificity VLDLR in the Xenopus oocyte with an estimated internalization rate of 2×10−3 s−1 (Opresko and Wiley, 1987). They further showed that the t1/2 of recovery of trypsin-digested VLDLR at the PM is ∼2 h. This indicates rapid recycling of the VLDLR at the PM (within minutes), and its replenishment through a large intracellular pool of VLDLR that supports the steady-state levels of VLDLR at the PM (Opresko and Wiley, 1987).
In this study, we extend the role of the VLDLR beyond lipoprotein uptake and signaling in the brain, to include for the first time a chaperone-trafficking function for mPR. We show that the VLDLR is required for the transit of mPR from the ER to the Golgi. The intracellular distribution of the VLDLR in mammalian cells is reminiscent of what we observe here in the oocyte, with both ER and Golgi localization (Wagner et al., 2013), suggesting a similar trafficking chaperone role in these cells. In the absence of VLDLR, mPR is enriched in the ER and is unable to traffic to the Golgi and then to the PM. Since mPR non-genomic signaling to release meiotic arrest requires its membrane residence, the VLDLR becomes essential for mPR-dependent signaling, as it is required for the trafficking and membrane residence of mPR. Knockdown of VLDLR phenocopies mPR inhibition in the oocyte and blocks P4-dependent release of oocyte meiotic arrest.
In conclusion, by using an unbiased quantitative proteomics approach, we identify the VLDLR as an mPR-interacting protein. We further show that the VLDLR is required for the trafficking and PM localization of mPR by mediating its transit from the ER to the Golgi. These results extend the known functions of this versatile receptor, to include a chaperone trafficking function for mPR. Given the broad tissue distribution and co-expression of VLDLR and mPRs, this raises the intriguing prospect that the VLDLR modulates P4-dependent non-genomic signaling in various physiological and pathological conditions. It is, as such, important to further explore the role of VLDLR in mPR-dependent non-genomic signaling, especially given the role of mPRs in cancer, reproductive regulation in both genders, and sugar homeostasis.
MATERIALS AND METHODS
Molecular biology
Coding sequences for Xenopus mPRβ (NCBI Reference Sequence: NM_001085861.1) and Xenopus VLDLR (GenBank: BC070552.1) were synthetized including or not including sequences for GFP (for mPR) and mCherry (Ch) (for VLDLR) at the N- or C-terminus. These were then cloned in pSGEM by Mutagenex Inc. pSGEM-TMEM-mCherry has been previously described (El-Jouni et al., 2007; Yu et al., 2009). GalNac–GFP was obtained from Brian Storrie at the University of Arkansas, Fayetteville, AR (Storrie et al., 1998), and subcloned into pSGEM. RNAs for all the clones were produced by in vitro transcription after linearizing the vectors with NheI using the mMessage mMachine T7 kit (Ambion). For VLDLR knock down, the sense and antisense oligonucleotides sequences used were as follows: sense, 5′-GATGGGAGTGTGATGGAGAC-3′; antisense, 5′-GTCTCCATCACACTCCCATC-3′. Relative expression of VLDLR and mPR were assessed by quantitative real-time PCR (qRT-PCR). Concomitant quantification of Xenopus ornithine decarboxylase mRNA transcripts (xODC) were used to normalize mRNA transcript levels (Šindelka et al., 2006). The forward (F) and reverse (R) primer sequences are as follows: VLDLR: F, 5′-CGATGGGAGTGTGATGGAG-3′; R, 5′-CTGCATTTGCACAGTCACG; mPR: F, 5′-CCTGTTGTCCACCGGATAGT-3′; R, 5′-GGTGACCGTGCCCTATAAAA-3′; xODC: F, 5′-GCCATTGTGAAGACTCTCTCCATTC-3′, R, 5′-TTCGGGTGATTCCTTGCCAC-3′.
Xenopus oocyte preparation and protein expression
Stage VI Xenopus laevis oocytes were obtained as previously described (Machaca and Haun, 2002). Animals were handled according to Weill Cornell Medicine College IACUC approved procedures (protocol #2011-0035). The oocytes were used 24 to 72 h after harvesting. Oocytes were injected with RNAs or sense or antisense oligonucleotides and kept at 18°C for 1–3 days after injection to allow for protein expression or efficient RNA degradation. After progesterone treatment, GVBD was recorded on a dissecting microscope by the appearance of a white spot at the animal pole. For the staining with WGA and P4–BSA–Fluorescein, oocytes that were completely denuded from attached follicular cells were selected by negative staining with Hoescht 33342 (Life Technologies) and were used for the P4–BSA–Fluorescein staining experiments.
Oocytes crosslinking and immunoprecipitation
Oocytes were incubated for 1 h at 4°C in freshly made ice-cold 10 mM NHS-acetate in crosslinking buffer (10 mM HEPES pH 8.0, 150 mM NaCl, 2 mM MgCl2), and then transferred to ice-cold freshly made 1 mM dithiobis(succinimidyl propionate) in crosslinking buffer for 30 min at 4°C. The crosslinking reaction was stopped by incubating the oocyte for 5 min in binding solution containing 5 mM HEPES, pH 7.6, 5 mM Tris-HCl, 100 mM NaCl, 2 mM KCl and 2 mM MgCl2. Oocytes were then lysed in IP solution (30 mM HEPES, 100 mM NaCl, pH 7.5) containing protease and phosphatase inhibitors (5 µl/oocyte). Lysates were cleared of yolk by centrifugation at 1000 g three times for 10 min each at 4°C. For the proteomics studies, lysates were then immunoprecipitated with the anti-GFP microbeads (1 µl/oocyte) using the GFP isolation kit (MACS Miltenyi Biotec) as per the manufacturer's instructions. For the co-immunoprecipitation studies with GFP-tagged mPR and mCherry-tagged VLDLR, lysates were first solubilized with 4% NP40 at 4°C for 2 h, followed by 15 min centrifugation at 14,000 rpm (18,400 g) at 4°C before being subjected to immunoprecipitation with anti-GFP microbeads.
Western blots
Cells were dounced in MPF lysis buffer [80 mM β-glycerophosphate, 20 mM Hepes (pH 7.5), 15 mM MgCl2, 20 mM EGTA, 1 mM Na-Vanadate, 50 mM NaF, 1 mM DTT, 1 mM PMSF and 0.1% protease Inhibitor (Sigma)] and centrifuged twice at 1000 g for 10 min each at 4°C to remove yolk granules. Supernatants were run on 4–12% SDS-PAGE gels, transferred to polyvinylidene difluoride (PVDF) membranes (Millipore), blocked for 1 h at room temperature with 5% milk in TBS-T buffer (150 mM NaCl, 20 mM Tris-HCl pH 7.6 and 0.1% Tween 20) and then incubated overnight at 4°C in 3% BSA in TBS-T with one of the following primary antibodies: anti-GFP (1:1000, Living Colors, Cat# 632381), anti-mCherry and anti-RFP (1:1000, Living Colors, Cat# 632543; 1:1000, Abcam, Cat# ab62341, respectively), anti-tubulin (1:10,000, Cell Signaling, Cat# 3873), anti-phospho-MAPK (ERK1/2) (1:5000, Cell Signaling, Cat# 9106) and anti-phospho-Cdc2 (1:1000, Cell Signaling, Cat# 9111). Blots were washed three times with TBS-T and probed for 1 h with horseradish peroxidase (HRP)-conjugated secondary antibody (1:10,000; Jackson ImmunoResearch) (for mCherry and Cdc2), or for 1 h with IRDye® 800 and 680-conjugated secondary antibodies (1:10,000; for GFP, RFP, MAPK and tubulin). The western blots were visualized by measuring either the chemiluminescence intensity from the peroxidase using the Immun-Star-HRP Chemiluminescent system (Bio-Rad), or by quantitative analysis using the LiCor Odyssey Clx Infrared Imaging system.
Proteomics
Elutes obtained after GFP binding and immunoprecipitation were trypsin digested and dimethyl labeled, with peptides from the cytosolic GFP–mPR immunoprecipitation (denoted ‘Cy’) labeled with the light isotope and those from the plasma membrane mPR-GFP immunoprecipitation (denoted ‘PM’) labeled with the heavy isotope. The mentioned labeling resulted in an amino acid mass increase of 28 Da per primary amine on a peptide for the light isotope, whereas an incorporation of the heavy isotope results in a mass increase of 36 Da. Dimethyl-labeled peptides from both immunoprecipitations were then mixed together and analyzed by mass spectrometry, allowing quantitative measurement of the differential enrichment of each detected peptide in the IPs by analyzing the ratio of heavy to the light isotope (PM:Cy). The proteomics analysis was conducted by the WCM-Q proteomic core. The experiment was repeated three times.
mPR and VLDLR imaging and quantification
Confocal imaging of live oocytes was performed using a LSM710 microscope (Zeiss, Germany) fitted with a Plan Apo 63×/1.4 NA oil immersion objective. Z-stacks were taken in 0.5 µm sections using a 1 Airy unit pinhole aperture. Images were analyzed using ZEN 2008 (Zeiss) or ImageJ software. To measure the distribution of mPR and VLDLR at the cell membrane, TMEM–mCherry or WGA were used as membrane markers. For each oocyte, the percentage of mPR and VLDLR at the plasma membrane was calculated by analyzing the intensity of fluorescence distribution through a z-stack of images, where we conservatively used two focal planes below the peak of TMEM–mCherry fluorescence, or the low point of WGA fluorescence, as a reference to mark the end of the plasma membrane. The data was normalized to the average mPR membrane percentage from the control condition. For colocalization analysis, GFP, mCherry or WGA fluorescence intensities were plotted as a function of the corresponding z-stack sections (µm).
Oocyte P4–BSA–fluorescein staining and imaging
Oocytes were first injected with VLDLR sense or antisense nucleotides, and incubated for 24 or 48 h. Oocytes were fixed at 4°C for 1 h in Ringer (in mM: 96 NaCl, 2.5 KCl, 1.8 CaCl2, 2 MgCl2, 10 HEPES, pH 7.4) containing 10% ethanol and BSA (8×10−7 M). Oocytes were then incubated at 4°C for 2 h in Ringer with 10% ethanol containing P4–BSA–Fluorescein (8×10−7 M) (Sigma-Aldrich), followed by three washes for 5 min each in Ringer. Oocytes were then imaged using a Zeiss LSM710 microscope with a 25× objective with the pinhole fully open by exciting the fluorescein fluorophore with the 488 nm laser and using the same master gain. ImageJ was then used to calculate the fluorescence in a specific region of interest (ROI). ROIs of the same dimensions were used for all the oocytes. The average fluorescence of P4-BSA-FITC from uninjected oocytes was used to normalize the P4–BSA–FITC intensity in the different conditions.
Electrophysiology
The SOCE-induced Cl− currents were recorded using a standard two-electrode voltage-clamp recording technique. Recording electrodes were filled with 3 M KCl and coupled to a Geneclamp 500B controlled with pClamp 10.5 (Axon Instruments). Ca2+-activated Cl− currents were recorded using a previously described ‘triple jump’ protocol (Courjaret and Machaca, 2016). Membrane capacitance was monitored using the built-in routine from pClamp to measure the cell membrane parameters. The cells were continuously superfused with Ringer buffer during voltage-clamp experiments using a peristaltic pump.
Statistics
Values are given as means±s.e.m. Statistical analysis was performed when required by using a paired or unpaired Student's t-test or ANOVA test. P values are indicated as follows: *P<0.05, **P<0.01, ***P<0.001, and ns, not significant. Each experiment is repeated at least three times from three independent donor females.
Acknowledgements
We thank the Proteomic Core at Weill Cornell Medicine Qatar for its support. The Core is supported by the BMRP program funded by Qatar Foundation.
Footnotes
Author contributions
Conceptualization: N.N., K.M.; Methodology: N.N., M.D., R.C., R.H., J.G.; Validation: N.N.; Formal analysis: N.N., J.G., K.M.; Investigation: N.N., M.D., R.C., R.H., R.M.; Resources: K.M.; Data curation: N.N., K.M.; Writing - original draft: N.N.; Writing - review & editing: K.M.; Visualization: K.M.; Supervision: K.M.; Project administration: K.M.; Funding acquisition: K.M.
Funding
This work was funded by the Qatar National Research Fund (QNRF) (NPRP 7-709-3-195). Additional support for the authors comes from the Biomedical Research Program (BMRP) at Weill Cornell Medical College in Qatar, a program funded by the Qatar Foundation. The statements made herein are solely the responsibility of the authors.
References
Competing interests
The authors declare no competing or financial interests.