Dephosphorylation of MAP2D enhances its binding to vimentin in preovulatory ovarian granulosa cells

ABSTRACT

Preovulatory granulosa cells express the low-molecular-mass MAP2D variant of microtubule-associated protein 2 (MAP2). Activation of the luteinizing hormone choriogonadotropin receptor by human choriogonadotropin (hCG) promotes dephosphorylation of MAP2D on Thr256 and Thr259. We sought to evaluate the association of MAP2D with the cytoskeleton, and the effect of hCG on this association. MAP2D partially colocalized, as assessed by confocal immunofluorescence microscopy, with the vimentin intermediate filament and microtubule cytoskeletons in naive cells. In vitro binding studies showed that MAP2D bound directly to vimentin and β-tubulin. Phosphorylation of recombinant MAP2D on Thr256 and Thr259, which mimics the phosphorylation status of MAP2D in naive cells, reduces binding of MAP2D to vimentin and tubulin by two- and three-fold, respectively. PKA-dependent phosphorylation of vimentin (Ser32 and Ser38) promoted binding of vimentin to MAP2D and increased contraction of granulosa cells with reorganization of vimentin filaments and MAP2D from the periphery into a thickened layer surrounding the nucleus and into prominent cellular extensions. Chemical disruption of vimentin filament organization increased progesterone production. Taken together, these results suggest that hCG-stimulated dephosphorylation of MAP2D at Thr256 and Thr259, phosphorylation of vimentin at Ser38 and Ser72, and the resulting enhanced binding of MAP2D to vimentin might contribute to the progesterone synthetic response required for ovulation.

Summary: The progesterone synthetic response of preovulatory granulosa cells might be facilitated by dephosphorylation of MAP2D and phosphorylation of vimentin, resulting in reorganization of vimentin filaments.

INTRODUCTION

Microtubule-associated protein (MAP)2 was originally identified as an abundant protein in neuronal cells that co-purified with microtubules (Dehmelt and Halpain, 2004). The four isoforms of MAP2 are encoded by a single gene through alternative splicing to yield high-molecular-mass MAP2A and MAP2B, and low-molecular-mass MAP2C and MAP2D isoforms. MAP2 isoforms bind microtubules with sub-micromolar affinity (Roger et al., 2004) through three (MAP2C) or four (MAP2A, MAP2B and MAP2D) C-terminal 31-amino acid microtubule-binding domains (MTBDs) (Lewis et al., 1988). MAP2 functionally binds and stabilizes microtubules, reducing their dynamic instability and hence enhancing their length (Gamblin et al., 1996; Weisshaar et al., 1992; Weisshaar and Matus, 1993). MAP2 proteins can be phosphorylated on many Ser and Thr residues (reviewed in Sanchez et al., 2000). MAP2 phosphorylation generally reduces the binding affinity of MAP2 to microtubules (Ozer and Halpain, 2000; reviewed in Sanchez et al., 2000).

MAP2 proteins, and specifically MAP2C, also bind F-actin (Cunningham et al., 1997) through the MTBDs (Correas et al., 1990; Roger et al., 2004) and organize actin into closely packed bundles (Roger et al., 2004). MAP2C binds actin with sub-micromolar affinity and is characterized as exhibiting ‘robust’ actin-bundling activity, evidenced by the ability of one MAP2C per 20 actin monomers to promote actin bundling (Roger et al., 2004). Although phosphorylation of MAP2 generally inhibits its actin filament cross-linking activity (Selden and Pollard, 1983), phosphorylation of the conserved KXGS domain within the MTBD enhances the association of MAP2C with actin in peripheral membrane ruffles (Ozer and Halpain, 2000). MAP2 proteins have also been reported to bind to at least two of the five classes of intermediate filaments, namely vimentin and neurofilaments (Bloom et al., 1985; Bloom and Vallee, 1983; Falconer et al., 1989; Miyata et al., 1986; Papasozomenos et al., 1985), using a site distinct from the MTBD (Heimann et al., 1985). The binding affinity of MAP2 to the 70-kDa neurofilament subunit is also reported to be sub-micromolar (Heimann et al., 1985; Miyata et al., 1986), although the functional significance of this interaction is not known.

MAP2 proteins are also A-kinase anchoring proteins (AKAPs) that scaffold protein kinase A (PKA) and other signaling proteins (Carr et al., 1999, 1993; Rubino et al., 1989; Salvador et al., 2004; Theurkauf and Vallee, 1982). Thus, the phospho-regulation and resulting relocation of MAP2 proteins to different cellular fractions has the potential to redistribute the PKA signaling module to enhance the efficiency of PKA substrate phosphorylation.

The ∼80-kDa MAP2D isoform is induced in ovarian granulosa cells by follicle-stimulating hormone (FSH) and, unlike most other FSH gene targets (reviewed in Hunzicker-Dunn and Mayo, 2015), its expression persists into luteal cells (Carr et al., 1993; Salvador et al., 2004), which secrete high levels of progesterone to maintain pregnancy (Niswender et al., 2000). MAP2D is the only MAP2 isoform detected in rat granulosa cells (Salvador et al., 2004). The majority of MAP2D in preovulatory granulosa cells is phosphorylated, based on its migration on SDS-PAGE gels at 80 kDa rather than at 70 kDa with dephosphorylated MAP2D (Flynn et al., 2008). Addition to primary granulosa cells of an ovulatory concentration of luteinizing hormone, mimicked by the luteinizing hormone chorionic gonadotropin receptor agonist human choriogonadotropin (hCG), promotes rapid dephosphorylation of MAP2D on Thr256 and Thr259 in a PKA-dependent manner (Flynn et al., 2008). Surprisingly, hCG does not promote phosphorylation of the Ser residue within the KXGS motif of the MTBDs (Flynn et al., 2008), although these Ser residues can be phosphorylated by PKA in vitro using recombinant MAP2D (Flynn et al., 2008) and are recognized PKA targets in other cells (Ozer and Halpain, 2000).

Preovulatory granulosa cells in primary culture on fibronectin substratum appear fibroblastic, with long bundles of F-actin (Karlsson et al., 2010). hCG promotes the PKA-dependent dephosphorylation of the actin-depolymerizing factor cofilin on Ser3, resulting in cell rounding and the appearance of spindly processes that appear neuronal-like (Karlsson et al., 2010). These events are required for the progesterone synthetic response (Karlsson et al., 2010) necessary for ovulation (Lydon et al., 1995; Robker et al., 2000).

Based on the prominent role of MAP2 proteins in regulating the microtubule and microfilament cytoskeleton and, hence, cell shape and function in neuronal cells, we sought to evaluate the association of MAP2D in preovulatory granulosa cells with the cytoskeleton and the effect of hCG on this association. Confocal immunofluorescence microscopy results show that MAP2D partially colocalizes with the intermediate filament vimentin and microtubule cytoskeletons, but not with the microfilament cytoskeleton in untreated (naive) preovulatory granulosa cells. Binding studies show that MAP2D binds directly to vimentin and to β-tubulin, and that the in vitro phosphorylation of recombinant MAP2D on Thr256 and Thr259, which mimics the phosphorylation status of MAP2D in untreated granulosa cells, reduces binding of MAP2D to vimentin two-fold and to tubulin by three-fold. Activation of the luteinizing hormone choriogonadotropin receptor, a G-protein coupled receptor that drives dephosphorylation of MAP2D on Thr256 and Thr259, promotes rapid PKA-dependent phosphorylation of vimentin on Ser38 and Ser72, a coincident increase in the binding of vimentin to MAP2D (∼44%), and contraction of granulosa cells with coincident reorganization of vimentin filaments and MAP2D from the periphery into a thickened layer surrounding the nucleus and prominent cellular extensions. The ability of the vimentin networks to rapidly redistribute within epithelial granulosa cells in response to luteinizing hormone choriogonadotropin receptor activation is consistent with the recognized dynamic properties of intermediate filaments (reviewed in Helfand et al., 2003). Artificial disruption of vimentin filaments increased progesterone production in granulosa cells in the absence of trophic hormone by two-fold. These results suggest that in preovulatory granulosa cells, the hCG-stimulated, PKA-dependent dephosphorylation of MAP2D at Thr256 and Thr259, phosphorylation of vimentin at Ser38 and Ser72, and resulting enhanced binding of MAP2D to vimentin might contribute to the progesterone synthetic response required for ovulation.

RESULTS

MAP2D colocalizes with the vimentin intermediate filament and microtubule cytoskeletons in granulosa cells as assessed by confocal immunofluorescence microscopy

We initially sought to determine whether MAP2D, the majority of which is highly phosphorylated in naive granulosa cells (Flynn et al., 2008), colocalized with the microtubule, microfilament and/or intermediate cytoskeleton in granulosa cells. Untreated fixed granulosa cells appear flattened and fibroblastic with long bundles of phalloidin-stained F-actin (Karlsson et al., 2010). Staining with antibodies against β-tubulin, vimentin [the predominant intermediate protein in granulosa cells (Albertini and Kravit, 1981)], actin and MAP2D proteins also showed that each of these proteins reached into the periphery of the cells (Fig. 1A–C). Dual staining with antibodies against β-tubulin and MAP2D revealed that a portion of MAP2D localized to the microtubule cytoskeleton (Fig. 1A). Dual staining for vimentin and MAP2D demonstrated that MAP2D also appeared to colocalize to a portion of vimentin intermediate filaments especially in the peri-nuclear region (Fig. 1B, arrowheads). However, colocalization of vimentin and MAP2D in naive granulosa cells was variable (see top panel of Fig. 1D), even within granulosa cells in the same vision field. Comparisons between MAP2D and microfilaments, as determined by staining for β-actin, surprisingly revealed no colocalization (Fig. 1C). These results demonstrate that MAP2D partially localizes to both the microtubule and vimentin intermediate filament cytoskeletons in granulosa cells but not to microfilaments. As the colocalization of MAP2 proteins with microtubules is extremely well-documented, we focused on the association of MAP2D and vimentin.

Fig. 1.

MAP2D partially colocalizes with the vimentin intermediate filament and microtubule cytoskeletons in granulosa cells, and hCG stimulates reorganization of the vimentin cytoskeleton. Confocal microscopy was performed on granulosa cells plated on fibronectin-coated glass coverslips, fixed in methanol, and visualized by indirect immunofluorescence, as stated in the Materials and Methods. (A) Cells were stained with anti-MAP2 rabbit polyclonal antibody (red) and anti-β-tubulin mouse monoclonal antibody (green). (B) Cells were stained with anti-MAP2 rabbit polyclonal antibody (red) and vimentin mouse monoclonal antibody (green). Arrowheads indicate peri-nuclear colocalization of vimentin and MAP2D. (C) Cells were stained with anti-MAP2 rabbit polyclonal antibody (red) and anti-β-actin mouse monoclonal antibody (green). (D) Plated granulosa cells were left untreated (–) or treated with 1 IU/ml hCG for 30 min before being fixed in methanol and stained with antibodies against MAP2 (red) and vimentin (green). Clusters of granulosa cells were chosen for imaging, as these cells appeared most responsive to hormonal treatment. Results are representative of at least three separate experiments. Scale bars: 10 µm.

Treatment with 1 IU hCG/ml for 30 min results in the rounding of granulosa cells and the appearance of dendritic-like cellular extensions (Karlsson et al., 2010). F-actin is reorganized from long fibers into a nondescript punctate distribution within both the cell body and extensions of granulosa cells (Karlsson et al., 2010). Consistent with our previous results (Karlsson et al., 2010), dual staining of cells with antibodies to vimentin and MAP2D showed that hCG promoted the contraction of granulosa cells and appearance of long cellular extensions (Fig. 1D). Vimentin filaments were redistributed mostly to a thickened layer surrounding the nucleus and to the cellular extensions. MAP2D, which is dephosphorylated at Thr256 and Thr259 in response to hCG (Flynn et al., 2008), was colocalized to the vimentin filaments at both cellular locations. These results show that granulosa cells contract as vimentin filaments and MAP2D are reorganized from the periphery into a thickened layer surrounding the nucleus and into cellular extensions in response to luteinizing hormone choriogonadotropin receptor activation.

MAP2D and vimentin localization is dependent on microtubule stability in granulosa cells

As vimentin particles largely rely on microtubule-dependent transport by molecular motors for the formation of mature intermediate filaments (reviewed in Helfand et al., 2004), vimentin networks are often assembled in close proximity to the microtubule cytoskeleton (Albertini and Clark, 1981; Gurland and Gundersen, 1995). This dependence on microtubules is such that microtubule destabilization often results in reorganization or even collapse of intermediate filaments into thickened ‘peri-nuclear cables’ (Albertini and Clark, 1981; Albertini and Kravit, 1981; Bloom and Vallee, 1983; Gurland and Gundersen, 1995; Hynes and Destree, 1978; Kumar et al., 2007; Starger and Goldman, 1977). To evaluate the importance of microtubule stability for vimentin and MAP2D localization in naive granulosa cells, tubulin and vimentin were observed by dual staining confocal immunofluorescence microscopy in cells treated without and with the microtubule-destabilizing drug nocodazole. In untreated cells, vimentin and microtubule cytoskeletal networks were often found colocalized, particularly in the peri-nuclear region (Fig. 2A). Upon disruption of the microtubule cytoskeleton by treatment with 1 µg/ml nocodazole for 60 min (Fig. 2B,C), vimentin networks became unorganized, disappearing from the periphery of cells and collapsing into a thickened layer surrounding the nucleus, as has been described previously (Albertini and Kravit, 1981). MAP2D was also localized to this region and remained colocalized with vimentin. These results confirm that MAP2D and vimentin are colocalized and their localization is dependent on microtubule stability.

Fig. 2.

MAP2D and vimentin localization is dependent on microtubule stability in granulosa cells. Confocal microscopy was performed on granulosa cells as described in Fig. 1. (A) Untreated cells were fixed and stained with antibodies against vimentin (red; rabbit polyclonal) and β-tubulin (green). (B,C) Granulosa cells were treated with 1 µg/ml nocodazole for 60 min. (B) Cells were stained with antibodies against MAP2 (red; rabbit polyclonal) and β-tubulin (green). (C) Cells were stained with antibodies against MAP2 (red; rabbit polyclonal) and vimentin (green; mouse monoclonal). Results are representative of at least two separate experiments.

hCG promotes redistribution of MAP2D dephosphorylated at Thr256 and Thr259 to a detergent-insoluble cell fraction

We sought to confirm the colocalization of MAP2D and vimentin by an alternative approach. Granulosa cells, treated with or without hCG for 15 min in suspension, were subjected to a standard cytoskeletal fractionation protocol at 35°C that preserves the in vivo assembled microtubules in their polymerized state, maintains unpolymerized tubulin as a soluble protein and distributes intermediate filaments to a detergent insoluble fraction (Minotti et al., 1991; Vallee, 1982; Zackroff and Goldman, 1979). Results (Fig. 3A, lane 1) showed that the detergent-insoluble fraction contained the majority of the intermediate filament protein vimentin. The majority of β-tubulin was detected in the microtubule-enriched fraction (Fig. 3A, lane 3), and β-actin was detected in both soluble and microtubule-enriched fractions (Fig. 3A, lanes 2 and 3).

Fig. 3.

hCG enhances the co-isolation of MAP2D with vimentin. (A) Granulosa cells were left untreated (–) or treated with 1 IU/ml hCG for 15 min in suspension before lysis at 37°C. The intermediate-filament-enriched detergent-insoluble (DI) fraction was pelleted by a low-speed centrifugation. Supernatants were centrifuged at 100,000 g (35°C) to separate intact microtubule networks and associated proteins (MT) from soluble protein (Sol), as detailed in the Materials and Methods. The soluble fraction contains 58±10% (mean±s.e.m., n=5) of MAP2D phosphorylated at Thr256 and Thr259. Results are representative of five separate experiments. (B–D) The total MAP2D levels in the DI, Sol and MT fractions of untreated (–) and hCG-treated granulosa cells were quantified by densitometric analysis of western blots. Values are for MAP2D levels are given as a percentage of total MAP2D for each treatment condition, as indicated. Values are the mean±s.e.m. from five separate experiments. In E, the change in the percentage of MAP2D in the DI and Sol fractions after hCG treatment (from B and C) is shown. *P<0.05 (unpaired Student's t-test).

The bulk of MAP2D phosphorylated at Thr256 and Thr259 was distributed to the soluble fraction (Fig. 3A, lane 2) of untreated granulosa cells. Treatment with hCG resulted in the dephosphorylation of MAP2D at Thr256 and Thr259 (Fig. 3A, lanes 4–6), consistent with previous results (Flynn et al., 2008). Despite the presence of saturating concentrations of phosphatase inhibitors, at least half of the MAP2D migrated at 70 kDa (Fig. 3A, lanes 1 and 4) in a hypo-phosphorylated state (Flynn et al., 2008). This result likely reflects dephosphorylation of MAP2D at select sites during isolation of the microtubule cell fraction (at 35°C), as (1) omission of phosphatase inhibitors during this isolation protocol resulted in migration of 100% of the MAP2D at 70 kDa, and (2) 89% of MAP2D in cell lysates collected at 4°C migrates at 80 kDa and reverts to the 70 kDa size upon incubation with lambda phosphatase (Flynn et al., 2008).

The majority of total MAP2D (74.8±3.6%; mean±s.e.m., n=5) in untreated granulosa cells was localized to the detergent-insoluble fraction, whereas 11.3±2.5% was detected in the soluble fraction (Fig. 3A–C). hCG promoted a reduction in total MAP2D detected in the soluble fraction (Fig. 3A, compare lanes 2 and 5, from 11.3±2.5% in untreated to 5.2±2.3% in hCG-treated granulosa cells, as a percentage of total MAP2D detected in all three fractions; mean±s.e.m., n=5). We consistently detected a corresponding (8.9±3.7%) increase in MAP2D in the detergent-insoluble fraction in hCG-treated cells (Fig. 3B, from 74.8±3.6% to 83.7±2.8%; n=5) whereas the distribution of MAP2D to the microtubule-enriched fraction (Fig. 3D) was not altered by hCG treatment (from 13.9±3.4% to 11.1±3.0%; n=5).

When results are calculated as the change in the percentage of MAP2D in the detergent-insoluble and soluble fractions in hCG-treated cells (Fig. 3E), results show that a pool of MAP2D redistributes from the soluble to the detergent-insoluble vimentin-enriched fraction upon dephosphorylation of MAP2D at Thr256 and Thr259 mediated by luteinizing hormone choriogonadotropin receptor activation (P<0.05).

Granulosa cell MAP2D binds to recombinant vimentin in a phosphorylation-dependent manner

Colocalization of MAP2D and vimentin as assessed by confocal immunofluorescence microscopy and cytoskeletal fractionation are consistent with the idea, but do not prove that, MAP2D binds to vimentin. We thus sought to determine whether a pool of MAP2D binds directly to a pool of vimentin and whether this association is enhanced upon hCG-stimulated dephosphorylation of MAP2D at Thr256 and Thr259, as suggested in Figs 1 and 3. Solid-phase overlay assays were performed using granulosa cell lysates containing MAP2D in various states of Thr256 and Thr259 phosphorylation. Clarified granulosa cell lysates were incubated for 30 min at 4°C or 30°C with or without 0.2 µM okadaic acid, a preferential inhibitor of protein phosphatase 2A (PP2A) (Favre et al., 1997; Resjo et al., 1999). Incubation of lysates at 4°C maintained phosphorylation of MAP2D at Thr256 and Thr259 (Fig. 4A, lane 1) whereas incubation at 30°C resulted in dephosphorylation of MAP2D at Thr256 and Thr259 (lane 2) mediated by endogenous phosphatase activity. Inhibition of endogenous PP2A activity by the addition of 0.2 µM okadaic acid was sufficient to block this dephosphorylation of Thr256 and Thr259 at 30°C (lane 3). Solid-phase overlay assays were then performed (Heimann et al., 1985). Granulosa cell lysates from Fig. 4A were incubated with purified tubulin, vimentin and actin immobilized on nitrocellulose membranes, and MAP2D bound to the immobilized protein was detected by use of an anti-MAP2 polyclonal antibody (Fig. 4B). MAP2D that was phosphorylated at Thr256 and Thr259 (incubated at 4°C, or 30°C with okadaic acid) bound in detectable amounts to vimentin and tubulin, but only weakly to actin. No change in actin binding was observed for MAP2D dephosphorylated at Thr256 and Thr259 (incubated at 30°C), and only a small increase in tubulin binding was observed. In comparison, binding between vimentin and dephosphorylated MAP2D (incubated at 30°C) was ∼four-fold greater (P<0.05) than for phosphorylated MAP2D (incubated at 30°C with okadaic acid) (Fig. 4B,C). No binding was detected when the assay was performed using cell-free buffer only, with no primary antibody or with an irrelevant primary antibody (not shown). These findings confirm that granulosa cell MAP2D binds, directly or indirectly, to vimentin and tubulin, and that this binding is enhanced upon dephosphorylation of MAP2D at Thr256 and Thr259.

Fig. 4.

MAP2D binds to vimentin in a phosphorylation-dependent manner. Plated granulosa cells were lysed by sonication and centrifuged to remove insoluble material. Soluble extracts were incubated at 4°C or 30°C with or without 0.2 µM okadaic acid (OA), as indicated. (A) Western blots were performed post-incubation with indicated antibodies to confirm the phosphorylation state of MAP2D. (B) 5 µg of the cytoskeletal proteins tubulin (T; purified bovine brain), vimentin (V; recombinant) and actin (A; purified rabbit muscle) were electrophoretically separated and transferred to membrane. Membranes were incubated in soluble granulosa cell extracts from A, washed and incubated with anti-MAP2 polyclonal antibody followed by secondary antibody and ECL to detect MAP2D bound to immobilized cytoskeletal proteins. Total amounts of membrane-bound cytoskeletal proteins were verified by staining with Amido Black. Results are representative of three separate experiments. (C) vimentin-bound MAP2D levels were quantified by densitometric analysis, and normalized to total vimentin protein levels as determined by Amido Black staining. Values are the mean±s.e.m. from three separate experiments. *P<0.05 (unpaired Student's t-test).

In vitro phosphorylation of recombinant MAP2D at Thr256 and Thr259 reduces direct binding between MAP2D and vimentin

We next determined whether MAP2D can bind directly to vimentin and whether phosphorylation of MAP2D at Thr256 and Thr259 modulates this interaction. Solid-phase overlays were performed using bacterially expressed and purified MAP2D labeled with Alexa Fluor 680 and incubated in vitro with no kinase added (Fig. 5A, lane 1), with the PKA catalytic subunit only (lane 2), with GSK3β only (lane 3) or with both kinases (lane 4). MAP2D in untreated granulosa cells is phosphorylated minimally at three of 15 potential phosphorylation sites (Ser136, Thr256 and Thr259) and is likely phosphorylated at additional sites by unidentified kinases, based on its migration in total cell extracts primarily at 80 kDa on SDS-PAGE gels (Flynn et al., 2008). GSK3β, a kinase that preferentially phosphorylates substrates previously phosphorylated by another kinase (Doble and Woodgett, 2003), maintains basal phosphorylation of MAP2D at Thr256 and Thr259 in untreated granulosa cells (Flynn et al., 2008).

Fig. 5.

In vitro GSK3β phosphorylation regulates direct binding between MAP2D and vimentin. Bacterially expressed and purified MAP2D was labeled with Alexa Fluor 680 and then subjected to sequential phosphorylation by PKA and GSK3β. (A) Western blots were performed post incubation to confirm the phosphorylation state of MAP2D. (B) RIIα, tubulin, vimentin and IgG were immobilized by spotting onto Immobilon-FL membrane and incubated overnight with Alexa-Fluor-680-labeled MAP2D that was not phosphorylated (left panel), only phosphorylated by PKA (middle column, +PKA), or phosphorylated by PKA and GSK3β (right panel, +PKA/GSK3β) from A. After washing, binding of Alexa-Fluor-680-labeled MAP2D was detected by using the Odyssey Infrared Imaging System. Results in A and B are representative of three separate experiments. (C) Mean densities of MAP2D bound to each amount of immobilized RIIα, vimentin or tubulin protein were quantified (n=3, see Materials and Methods). Slopes of lines fitted to plotted data were calculated.

In vitro phosphorylation results with recombinant MAP2D show that sequential phosphorylations with the PKA catalytic subunit and GSK3β (lane 4) were necessary to promote phosphorylation of MAP2D at Thr256 and Thr259. Although PKA does not appear to be the physiological kinase that primes MAP2D for phosphorylation by GSK3β (Flynn et al., 2008), PKA promoted phosphorylation of recombinant MAP2D at sites recognized by the phospho-PKA substrate antibody (lane 2). We thus used MAP2D phosphorylated with PKA plus GSK3β to promote phosphorylation at Thr256 and Thr259 to mimic the phosphorylation status of MAP2D Thr256 and Thr259 in naive granulosa cells. We used MAP2D phosphorylated only with PKA to mimic MAP2D that is not phosphorylated at Thr256 and Thr259 and thus to reflect the phosphorylation status of MAP2D Thr256 and Thr259 in hCG-treated granulosa cells.

hCG does not stimulate the phosphorylation of MAP2D on PKA sites in intact granulosa cells, based on the absence of phospho-PKA substrate antibody signal in MAP2D immunoprecipitated from vehicle versus hCG-treated cells (Flynn et al., 2008). However, PKA does promote the phosphorylation of recombinant MAP2D at Ser residues within KXGS motifs within the MTBDs, as detected using a phospho-antibody (12E8) to the Ser residue within this motif (Flynn et al., 2008). Although Ser136 on MAP2D is phosphorylated in naive granulosa cells, this residue is not a PKA phosphorylation site (Berling et al., 1994).

Solid-phase overlay assays were performed by incubating Alexa-Fluor-680-labeled MAP2D that was not phosphorylated (MAP2D), MAP2D phosphorylated by PKA (MAP2D+PKA), or MAP2D phosphorylated by PKA and GSK3β (MAPD+PKA/GSK3β) with indicated amounts of purified PKA regulatory subunit IIα (also known as PRKAR2A, hereafter denoted RIIα), vimentin, tubulin or IgG immobilized on Immobilon-FL membranes. Fig. 5B is a representative blot showing the binding of recombinant MAP2D to RIIα, vimentin, tubulin and IgG; these data are quantitatively depicted in Fig. 5C. Results show that the density of bound MAP2D increased in a linear fashion with increasing amounts of immobilized RIIα, vimentin, and tubulin. Slopes of lines fit to the data were used to compare the relative binding affinities of MAP2D to immobilized proteins. Non-phosphorylated MAP2D bound to RIIα, vimentin and tubulin with the highest relative affinity (Fig. 5C, black diamonds). However, MAP2D always appears to be present in neuronal cells (Sanchez et al., 2000) and granulosa cells (Flynn et al., 2008; Salvador et al., 2004) in a phosphorylated state.

A comparison of the line slopes for the binding of phosphorylated MAP2D to vimentin, tubulin and RIIα shows that phosphorylation by PKA alone (likely on Ser within KXGS motifs; Flynn et al., 2008) decreased the relative affinity of MAP2D binding to RIIα by 1.67±0.03-fold (mean±s.e.m.; n=3), to vimentin by 4.86±1.14-fold (n=3), and to tubulin by 9.17±2.46-fold (n=3) (Fig. 5C). Phosphorylation by PKA+GSK3β, which promotes phosphorylation of Thr256 and Thr259, compared to PKA alone further decreased the relative affinity of MAP2D binding to vimentin by 1.91±0.07-fold (n=3) and to tubulin by 2.3±0.07-fold (n=3) but did not affect MAP2D binding to RIIα (0.93±0.09, n=3).

These results demonstrate that, similar to the well described interaction between MAP2 and tubulin (Lewis et al., 1988), MAP2D binds directly to vimentin. The relative affinity of this interaction is reduced by GSK3β phosphorylation at sites including Thr256 and Thr259. hCG-stimulated dephosphorylation of Thr256 and Thr259 in granulosa cells would thus be expected to enhance the relative binding affinity of MAP2D to vimentin.

As GSK3β might phosphorylate sites in addition to Thr256 and Thr259, we tested the importance of the phospho-regulation of MAP2D specifically at Thr256 and Thr259 on the relative binding of MAP2D to vimentin and tubulin. We compared the line slopes for the binding of wild-type MAP2D and the phospho-mimetic Glu256 and Glu259 mutant MAP2D. Both wild-type and mutant MAP2D were subjected to phosphorylation reactions without or with the PKA catalytic subunit (Fig. 6A). PKA promoted phosphorylation of recombinant wild-type and mutant MAP2D at sites recognized by the phospho-PKA substrate antibody (lanes 2 and 4). Consistent with results in Fig. 5C, quantification of dot blot binding of MAP2D to RIIα was not affected by phospho-regulation of Thr256 and Thr259 (Fig. 6B). However, mutation of MAP2D Thr256 and Thr259 to Glu256 and Glu259 reduced the relative binding of MAP2D compared to wild-type MAP2D to vimentin by 2.2±0.3-fold (mean±s.e.m., n=3) and to tubulin by 3.3±0.06-fold (n=3) (Fig. 6C,D). Mutation of these sites also reduced binding of PKA-phosphorylated MAP2D to both vimentin (2.2±0.3-fold, n=3) and tubulin (3.56±0.47-fold, n=3).

Fig. 6.

The phospho-mimetic Glu256 and Glu259 mutant MAP2D binds to vimentin with lower affinity than wild-type MAP2D. Alexa-Fluor-680-labeled wild-type (WT) and Glu256 Glu259 mutant MAP2D [MAP2D(E256/E259)] were subjected to in vitro phosphorylation reactions without or with the catalytic subunit of PKA, described in the Materials and Methods. (A) Western blots were performed post incubation to confirm phosphorylation state of MAP2D. (B–D) Slopes of lines fitted to plotted data for MAP2D bound to immobilized RIIα, vimentin and tubulin were calculated, as described in the Fig. 5 legend.

Taken together, these results show that the binding of MAP2D not only to tubulin (reviewed in Sanchez et al., 2000) but also to vimentin is negatively affected by the phosphorylation status of MAP2D. However, PKA does not appear to phosphorylate MAP2D in granulosa cells (Flynn et al., 2008); hence, the reduced binding affinity of MAP2D to vimentin or tubulin promoted by PKA phosphorylation in granulosa cells is unlikely to be physiologically relevant. In contrast, the physiologically relevant phosphorylation specifically of Thr256 and Thr259 reduces the relative binding affinity of MAP2D for vimentin by 2.2-fold.

hCG promotes PKA-dependent phosphorylation of vimentin, enhanced binding of vimentin to MAP2D and redistribution of the vimentin cytoskeleton

As vimentin is phosphorylated in vitro and in vivo at Ser38 and Ser72 by PKA (Ando et al., 1989; Eriksson et al., 2004; Geisler et al., 1989), the apparent hCG-dependent redistribution of the AKAP MAP2D dephosphorylated at Thr256 and Thr259 into a cell fraction containing vimentin (Fig. 3E) prompted us to investigate the phosphorylation of vimentin at these PKA sites in granulosa cells. Treatment of granulosa cells with hCG for 10 or 30 min resulted in increased vimentin phosphorylation at Ser38 and Ser72 (Fig. 7A, lanes 2 and 3); pretreatment with the PKA inhibitor (Cheng et al., 1986) myristoylated-PKI (Myr-PKI) blocked the hCG-induced increase in vimentin phosphorylation at Ser38 (Fig. 7B, lanes 2 and 4). These results indicate that, coincident with hCG-stimulated PKA-dependent dephosphorylation of MAP2D at Thr256 and Thr259 (Flynn et al., 2008) and its apparent redistribution to a cell fraction containing vimentin (see Figs 1 and 3), vimentin is phosphorylated in a PKA-dependent manner at recognized PKA sites.

Fig. 7.

hCG promotes vimentin phosphorylation at PKA sites. (A) Granulosa cells were left untreated (–) or treated with 1 IU/ml hCG for the indicated times. (B) Cells were left untreated or pretreated with 50 µM Myr-PKI for 60 min and then left untreated (–) or treated with 1 IU/ml hCG for 10 min. (C) Unpolymerized vimentin was incubated without or with the catalytic subunit of PKA. The phosphorylation state of vimentin was confirmed by western blotting with anti-phospho-PKA substrate antibody. Results in A–C are representative of three separate experiments. (D) The relative binding affinity of Alexa-Fluor-680-labeled MAP2D to vimentin that was either not phosphorylated or phosphorylated by PKA was determined, as described in the Fig. 5 legend.

We next investigated whether phosphorylation of vimentin at PKA sites affected its relative binding to MAP2D. We compared the relative binding affinity of Alexa-Fluor-680-labeled MAP2D with vimentin that was either not phosphorylated (Fig. 7C, lane 1) or phosphorylated by PKA (lane 2). Quantification of dot blot binding studies (Fig. 7D) showed that phosphorylation of vimentin at PKA sites enhances binding of vimentin to MAP2D by 44.2±7.4% (mean±s.e.m.; P<0.05, n=4).

It is well established that intermediate filaments are dynamic structures (reviewed in Helfand et al., 2004), and are important for changes in cell shape and morphology (Goldman et al., 1996). Furthermore, it has been shown that turnover of vimentin filaments is regulated by phosphorylation (Ando et al., 1989; Eriksson et al., 2004; Evans, 1988; Inagaki et al., 1987; Lamb et al., 1989). In particular, phosphorylation of vimentin at Ser38 and Ser72 by PKA has been linked to disassembly of vimentin filaments in vivo (Eriksson et al., 2004).

Disruption of vimentin results in increased progesterone production in granulosa cells

In view of the observed changes in granulosa cell morphology and apparent vimentin reorganization following treatment with hCG (see Fig. 1D) coupled with enhanced binding affinity of vimentin to MAP2D dephosphorylated at Thr256 and Thr259 (see Figs 5 and 6), we sought to evaluate the possible functional consequences of these cytoskeletal changes initiated by luteinizing hormone choriogonadotropin receptor signaling.

A dynamic vimentin cytoskeleton is thought to play a unique role in steroidogenic cells by modulating the accessibility of cholesterol-ester-containing lipid droplets to mitochondria (reviewed in Hall and Almahbobi, 1997; Shen et al., 2016). Lipid droplets from adrenal cells and adipocytes are surrounded by a vimentin-enriched phospholipid monolayer (Almahbobi et al., 1992; Brasaemle et al., 2004; Franke et al., 1987; Lieber and Evans, 1996). A recent report comparing the proteome of cholesterol-ester- versus triacylglycerol-enriched lipid droplets from rat granulosa cells demonstrated the enrichment of a number of proteins in cholesterol-ester-enriched lipid droplets that included vimentin (4.92-fold enrichment; Khor et al., 2014). Moreover, the genetic ablation of vimentin resulted in a 70% decrease in hCG-stimulated progesterone production, compared to wild-type mice, by preovulatory granulosa cells (Shen et al., 2012), although the mice remain fertile (Colucci-Guyon et al., 1994). Trophic hormones, through PKA (Rae et al., 1979), promote mobilization of free cholesterol from intracellular lipid droplets to the outer mitochondrial membrane (Liscum and Munn, 1999) by a series of events that appear to include the disassembly of vimentin filaments (Hall and Almahbobi, 1997; Lee and Mrotek, 1984; Shiver et al., 1992).

We therefore determined whether disruption of the vimentin intermediate filament network affected granulosa cell progesterone production. Granulosa cells placed on glass coverslips were treated without or with IDPN for 60 min. IDPN is a reversible intermediate filament disrupting agent (Galigniana et al., 1998; Griffin et al., 1978; Kumar et al., 2007; Lee and Mrotek, 1984; Papasozomenos et al., 1985) that selectively disrupts vimentin filament assembly without affecting actin or tubulin assembly (Kumar et al., 2007). This treatment destabilized vimentin filaments, as they were observed in a thickened band surrounding the nucleus (Fig. 8A). Intermediate filament destabilization resulted in characteristic cell rounding but had no apparent effect on microtubule stability, in agreement with observations of others (Kumar et al., 2007; Shiver et al., 1992). We then tested whether treatment of granulosa cells with IDPN modulated progesterone secretion into the culture medium by cells treated with or without hCG for 5 h. Results (Fig. 8B, open bars) showed that in the absence of hCG, IDPN promoted a 2.24-fold increase in progesterone secretion (P<0.05, n=3). In the presence of hCG, IDPN also significantly enhanced progesterone secretion (P<0.05, n=3).

Fig. 8.

Vimentin disruption stimulates increased progesterone secretion by granulosa cells. (A) Granulosa cells plated on glass coverslips were left untreated (–) or treated for 60 min with 1% (v/v) IDPN. Cells were stained with antibodies against vimentin (red; rabbit polyclonal) and β-tubulin (green) to confirm disruption of the vimentin cytoskeleton by IDPN treatment. (B) Granulosa cells plated on fibronectin were untreated or treated with 1% IDPN without or with 1 IU/ml hCG. After 5 h, media was collected and progesterone levels were measured by RIA. Results are means±s.e.m. of three separate experiments. Medium progesterone levels from cells treated without or with IDPN and not treated with hCG are expanded in right panel. Unpaired Student's t-test was used to compare vehicle versus hCG, vehicle versus vehicle, and hCG versus hCG treatments without and with IDPN. Different letters (a, b, c, d) refer to statistically different values (P<0.05).

Taken together, these results are consistent with the premise that hCG-stimulated movement of MAP2D dephosphorylated on Thr256 and Thr259 into the vimentin-enriched fraction, coupled with phosphorylation of vimentin on Ser38 and Ser72, facilitates vimentin disassembly that is necessary for progesterone production.

DISCUSSION

Formation of the preovulatory ovarian follicle capable of responding to the surge of luteinizing hormone with ovulation is characterized by expression of the AKAP MAP2D in granulosa cells (Carr et al., 1993; Salvador et al., 2004). MAP2D is highly phosphorylated in untreated granulosa cells and selectively dephosphorylated on Thr256 and Thr259 in response to the luteinizing hormone choriogonadotropin receptor agonist hCG (Flynn et al., 2008). Results herein show that in untreated granulosa cells, there is some colocalization of MAP2D not only with microtubules throughout the cytosol but also, albeit more variably, with the intermediate filament protein vimentin at a peri-nuclear location. hCG causes rapid rounding of granulosa cells and formation of long cellular extensions coincident with reorganization of both the vimentin network and MAP2D from the cell periphery to a thickened peri-nuclear layer and into cytoplasmic extensions. MAP2D that is not phosphorylated at Thr256 and Thr259 binds directly to vimentin with a higher relative affinity than MAP2D phosphorylated at these sites. Moreover, we show that vimentin is phosphorylated on Ser38 and Ser72 in a PKA-dependent manner upon luteinizing hormone choriogonadotropin receptor activation, and PKA-phosphorylated vimentin preferentially binds MAP2D compared to vimentin not phosphorylated by PKA. Consistent with the recognized ability of PKA-phosphorylated vimentin to promote disassembly of vimentin filaments (Eriksson et al., 2004) and our results showing that luteinizing hormone choriogonadotropin receptor activation leads to reorganization of vimentin filaments in granulosa cells and enhanced binding of MAP2D to vimentin in vitro, we asked whether these coincident events were physiologically relevant to granulosa cell function. Our results suggest that these responses to PKA activation might contribute to the hCG-dependent progesterone production obligatory for ovulation.

Consistent with our direct binding data (see Figs 5 and 6), a relatively small pool of MAP2D that is dephosphorylated on Thr256 and Thr259 in response to luteinizing hormone choriogonadotropin receptor activation redistributes from the soluble to the vimentin-enriched fraction upon cellular fractionation (see Fig. 3). Our direct binding results also suggest that this small pool of MAP2D dephosphorylated at Thr256 and Thr259 preferentially binds vimentin that is phosphorylated at Ser38 and Ser72 (see Fig. 7). However, the basis for the fractionation of the majority of MAP2D into the detergent-insoluble vimentin-enriched cellular fraction (see Fig. 3) is not supported by vimentin and MAP2D colocalization results from confocal immunofluorescence microscopy (see Figs 1 and 2). Thus, it is unlikely that the MAP2D detected in the detergent-insoluble fraction in untreated granulosa cells is a consequence of MAP2D binding to vimentin, especially given that MAP2D in untreated granulosa cells is highly phosphorylated (Flynn et al., 2008) and hence exhibits a reduced affinity for binding to vimentin (see Figs 5 and 6).

MAP2D dephosphorylated at Thr256 and Thr259 binds with a relatively higher affinity to vimentin than MAP2D phosphorylated at these sites. This direct evidence was obtained using bacterially expressed MAP2D that was either phosphorylated in vitro or mutated to produce a Thr256 and Thr259 phospho-mimetic. The caveat to our interpretation is that we do not know the extent of phosphorylation of MAP2D in granulosa cells, the kinases responsible for these phosphorylations, or the sequential order of these potential phosphorylations. Whereas our results indicate that Ser136, Thr256 and Thr259 are phosphorylated in untreated cells, it is likely that additional sites are phosphorylated because MAP2D migrates primarily at 80 kDa in total cell extracts from hCG-treated cells (in which Thr256 and Thr259 are dephosphorylated) (Flynn et al., 2008). Therefore, whereas selective modification of Thr256 and Thr259 clearly modulates the relative affinity by which vimentin and MAP2D interact, we cannot speak to the effect of other phosphorylation sites. However, it is clear from results presented in Figs 4–6 that phosphorylation of MAP2D generally reduces binding to vimentin.

Vimentin is the predominant intermediate filament protein in granulosa cells (Albertini and Kravit, 1981). Phosphorylation of vimentin by PKA is recognized to modify intermediate filament protein organization, often collapsing the filaments into tight bundles (Ando et al., 1989; Eriksson et al., 2004; Geisler et al., 1989). In particular, the in vivo phosphorylation of vimentin at Ser38 and Ser72 enhances vimentin dynamics by increasing the extent of vimentin disassembly (Eriksson et al., 2004). In light of the well-established role of MAP2D as an AKAP (Carr et al., 1993; Obar et al., 1989; Salvador et al., 2004) and our data suggesting the recruitment of a dynamic pool of this AKAP into the vimentin-enriched fraction upon luteinizing hormone choriogonadotropin receptor activation, we hypothesized that MAP2D dephosphorylation at Thr256 and Thr259 might function to enhance efficient PKA activity towards vimentin at Ser38 and Ser72, resulting in remodeling of the intermediate filament cytoskeleton. Although our results demonstrate that luteinizing hormone choriogonadotropin receptor signaling leads to phosphorylation of vimentin at these crucial residues, and that granulosa cells adopt a contracted cell shape and vimentin filaments are reorganized away from the periphery into the region surrounding the nucleus, we have not directly linked the movement of MAP2D dephosphorylated at Thr256 and Thr259 into the vimentin-enriched fraction to the vimentin phosphorylation and filament reorganization. However, based on these observations, we hypothesize that localization of MAP2D to vimentin filaments allows for more efficient vimentin phosphorylation by PKA at Ser38 and Ser72, resulting in an imbalance between filament assembly and disassembly and subsequent changes in the granulosa cell cytoskeleton.

One of the most prominent actions of the mid-cycle surge of luteinizing hormone is the induction of progesterone biosynthesis. Progesterone-receptor-null mice do not ovulate (Lydon et al., 1995) and blockade of luteinizing-hormone-stimulated progesterone biosynthesis with aminoglutethimide blocks ovulation (Szabo et al., 1996). Cholesterol, the substrate for steroid hormone biosynthesis, is stored in steroidogenic cells as cholesterol esters in lipid droplets (reviewed in Liscum and Munn, 1999). Upon stimulation by trophic hormones, cholesterol is de-esterified by hormone-sensitive lipase, transported to the outer mitochondrial membrane (Rae et al., 1979), and then moves to the inner mitochondrial membrane by the activity of steroidogenic acute regulatory protein, where it is cleaved to pregnenolone (reviewed in Miller, 2007; Stocco, 2000).

The cellular mechanisms that mediate the transfer of cholesterol from lipid droplets to the outer mitochondrial membrane are poorly understood. However, vimentin filaments are believed to play a unique role (reviewed in Hall and Almahbobi, 1997; Shen et al., 2016). Lipid droplets in steroidogenic cells are composed of a cholesterol ester core surrounded by a phospholipid monolayer that contains a large number of proteins, including vimentin and hormone sensitive lipase (reviewed in Shen et al., 2016). Vimentin has been shown to interact with both hormone-sensitive lipase and steroidogenic acute regulatory protein in adipocytes and adrenal cells, respectively (Shen et al., 2010, 2003). Vimentin-null mice exhibit a 70% reduction in progesterone production following pregnant mare's serum gonadotropin (PMSG) and hCG priming, which has been attributed to a defect in the movement of cholesterol from the cytosol to mitochondria (Shen et al., 2012). Finally, vimentin disassembly is believed to be necessary for cholesterol trafficking to mitochondria (reviewed in Hall and Almahbobi, 1997). Consistent with this hypothesis, chemical disruption of vimentin filaments increases steroidogenesis in adrenal cells (Lee and Mrotek, 1984; Shiver et al., 1992). Our results showing enhanced progesterone release from granulosa cells in response to the vimentin-filament-disrupting agent IDPN are consistent with the notion that cholesterol mobilization is facilitated by disruption of the vimentin intermediate filament network. However, direct proof for the involvement of vimentin phosphorylation and MAP2D dephosphorylation in progesterone production requires additional studies.

Conclusions

Our results show that dephosphorylation of MAP2D at Thr256 and Thr259 coupled with phosphorylation of vimentin at PKA sites enhances binding of vimentin and MAP2D.

These results support the hypothesis that dephosphorylation of MAP2D at Thr256 and Thr259 serves to regulate the association of MAP2D and possibly PKA with vimentin. This association might facilitate efficient phosphorylation of vimentin at Ser38 and Ser72 by PKA, resulting in the observed restructuring of vimentin intermediate filaments after hCG treatment. The binding between MAP2D and vimentin and subsequent vimentin remodeling likely plays a role in the luteinizing-hormone-dependent progesterone production by granulosa cells that is essential for fertility.

MATERIALS AND METHODS

Materials

The following were purchased: rabbit skeletal muscle actin, recombinant Syrian hamster vimentin, recombinant tubulin and bovine brain tubulin from Cytoskeleton, Inc. (Denver, CO); hCG from Abraxis Pharmaceutical Products (Schaumburg, IL); pregnant mare's serum gonadotropin (PMSG), pepstatin A, leupeptin, benzamidine, phenylmethylsulfonyl fluoride (PMSF), soybean trypsin inhibitor (STI), and Amido Black from Sigma-Aldrich (St Louis, MO); antipain dihydrochloride, calpain inhibitor II, E-64 and aprotinin from Roche; myristoylated PKA inhibitor amide 14-22 (Myr-PKI) from EMD Biosciences, Calbiochem (La Jolla, CA); okadaic acid from Alexis Biochemicals (Lausen, Switzerland); human fibronectin from BD Biosciences (San Jose, CA); β,β′-iminodipropionitrile (IDPN) from Fisher Scientific, Acros Organics (Waltham, MA); Immobilon-FL membrane from Millipore (Billerica, MA); ECL reagents, Rainbow molecular mass markers, and Hybond-C Extra nitrocellulose membranes from Amersham Biosciences, GE Healthcare (Buckinghamshire, UK).

Antibodies

Antibody sources and dilutions are presented in Table S1.

Animals

Immature female Sprague-Dawley rats (Charles River Laboratories, Portage, MI) were maintained in accordance with the Animal Welfare Act by protocols approved by the Northwestern University and Washington State University Institutional ACUC Committees (Flynn et al., 2008). Rats were injected at 22 day of age with 25 IU PMSG to induce formation of preovulatory follicles.

Granulosa cell culture

Granulosa cells were isolated from preovulatory follicles of 24-day-old rats and used immediately or plated on fibronectin-coated plastic dishes or glass coverslips in DMEM/F12 serum-free medium supplemented with 10 nM 17 β-estradiol, 100 U/ml penicillin, and 100 μg/ml streptomycin (Flynn et al., 2008).

Electrophoresis and western blot analysis

Proteins in total cell extracts were separated by SDS-PAGE using 10% separating gels and electrophoretically transferred to Hybond-C Extra nitrocellulose membrane; membranes were subjected to western blotting (Flynn et al., 2008). Films were scanned with an Epson 1640SU scanner and Adobe Photoshop version 7.0 software with minimal processing. Relative protein quantities were calculated from densitometric measurements of band intensities using Molecular Analyst software (Bio-Rad Laboratories, Inc., Hercules, CA). Results were analyzed by Graphpad Prism^® version 6.01 and are presented as the mean±s.e.m. on scatter plots.

Confocal immunofluorescence microscopy

Treatments of granulosa cells cultured on fibronectin-coated glass coverslips were terminated by aspirating medium and rinsing cells once with PBS. Cells were fixed in −20°C methanol (5 min), rinsed in PBS (20 min, 21°C), blocked in 1 mg/ml normal goat serum (Jackson ImmunoResearch) (15 min, 37°C), and incubated in primary antibody (30 min, 37°C). Coverslips were washed once in 0.05% Tween-20 in PBS and twice in PBS before incubation (30 min, 37°C) with goat anti-mouse IgG or goat anti-rabbit IgG antibody conjugated to Alexa Fluor 488 or Alexa Fluor 568, respectively. Coverslips were washed as above, dried, and mounted with Gelvatol onto glass slides. Slides were analyzed at the Cell Imaging Facility, Northwestern University Feinberg School of Medicine, using a Zeiss LSM510 laser-scanning confocal microscope at 60× magnification. All primary antibodies used react with a single protein band at appropriate molecular mass on SDS-PAGE gels.

Protease inhibition

Protease Inhibitor Cocktail was added to various buffers such that final concentrations were 10 µg/ml pepstatin A, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 50 µg/ml STI, 25 mM benzamidine, 10 µg/ml E-64, 1 mM PMSF, 7 µg/ml calpain inhibitor II and 50 µg/ml antipain dihydrochloride.

Cytoskeletal fractionation

Granulosa cells were treated without or with hormone in suspension (without plating on fibronectin), collected and lysed at 37°C by sonication in microtubule-stabilizing buffer (50 mM PIPES pH 6.6, 0.5% Nonidet P-40, 2 mM EGTA, 1 mM MgCl₂, 2 mM dithiothreitol, 1 mM Na₃VO₄, 20 mM NaF, 2.5 mM Na₂H₂P₂O₇, and protease inhibitor cocktail). A detergent-insoluble fraction was pelleted by a low-speed centrifugation (9000 g, 10 min) at 35°C. Low speed supernatant was centrifuged (100,000 g, 30 min) at 35°C to pellet the polymerized microtubule-enriched fraction. The remaining supernatant contained unpolymerized microtubules and other soluble cell components. Samples were heat denatured in SDS-PAGE sample buffer.

Solid-phase overlay assays with granulosa cell MAP2D

A total of 5 µg of tubulin, vimentin, or actin were separated by SDS-PAGE, then electrophoretically transferred to nitrocellulose membrane. Membranes were blocked with overlay blocking buffer (5% BSA in 0.05 M Tris-HCl pH 7.4, 5 mM EDTA, 0.19 M NaCl, and 2.5% Triton X-100) for 3 h (Heimann et al., 1985). Granulosa cells were plated overnight and then lysed by sonication in overlay lysis buffer (12.5 mM Tris-HCl pH 7.2, 100 mM NaCl, 0.5% Nonidet P-40, 1 mM EGTA and Protease Inhibitor Cocktail). Lysates were centrifuged at 16,000 g (20 min at 4°C) and supernatants incubated 30 min at 4°C or 30°C, with or without addition of 0.2 µM okadaic acid. Phosphatase inhibitors were then added (1 mM Na₃VO₄, 20 mM NaF, 2.5 mM Na₂H₂P₂O₇, 0.2 µM okadaic acid) to stop endogenous phosphatase activity. Immobilized cytoskeletal proteins were incubated with granulosa cell extracts (1.5 h, 21°C) with rotation, washed three times in overlay wash buffer [12.5 mM Tris-HCl pH 7.2, 100 mM NaCl with 1 mM Na₃VO₄, 20 mM NaF, 2.5 mM Na₂H₂P₂O₇, 0.2 µM okadaic acid (added to first wash only)], and incubated (2 h, 21°C) with anti-MAP2 primary antibody (1:500 dilution of polyclonal antibody in 5% milk in TBS). Membranes were washed as before, incubated with secondary antibody (1 h, 21°C), washed again, and bound MAP2 protein–antibody complexes detected by enhanced chemiluminescence. Membranes were stained with Amido Black to visualize total protein.

Solid-phase overlay assays with recombinant MAP2D

Full-length rat MAP2D-pET30a was expressed in E. coli BL21 (DE3) pLysS (Novagen) by 1 mM IPTG (Sigma-Aldrich) induction and purified in a HiTrap Chelating HP column (GE Healthcare). To create a construct in which Thr256 and Thr259 were mutated to glutamic acid residues, site-directed mutagenesis was performed by using the QuikChange II site-directed mutagenesis kit from Stratagene (La Jolla, CA) with the following primers (mutated nucleotides are shown in italics): forward primer (5′-CAAGCTACTCTTCACGTGAACCAGGCGAACCTGGAACCCCGAGC-3′), reverse primer (5′-GCTCGGGGTTCCAGGTTCGCCTGGTTCACGTGAAGAGTAGCTTG-3′). Mutations were confirmed by sequence analysis, and purification was confirmed by western blotting. After purification, wild-type and Glu256 and Glu259 mutant MAP2D-pET30 were dialyzed into PBS, and then labeled with the Alexa Fluor 680 protein labeling kit, according to instructions (Molecular Probes, Eugene, OR). Purified, Alexa-Fluor-680-labeled wild-type and Glu256 and Glu259 mutant MAP2D-pET30 were incubated in kinase buffer (50 mM MOPS, 10 mM MgCl₂, 0.25 mg/ml BSA, pH 7.0) with 2 mM MgATP, and 0.22 µg PKA catalytic subunit (or water for non-phosphorylated control) for 20 min (21°C). Glycogen synthase kinase 3β (GSK3β; 2 µl, or water for non-phosphorylated and PKA-only controls) was then added and incubation was continued for 1 h (30°C). Purified recombinant murine PKA regulatory subunit IIα (RIIα)-pET11 (Carr et al., 1993), rabbit IgG, recombinant tubulin polymerized according to manufacturer's instructions and unpolymerized recombinant vimentin were spotted onto the Immobilon-FL membrane.

For overlays in which the effect of PKA-mediated phosphorylation of vimentin was investigated, phosphorylation reactions were performed as above, replacing MAP2D with 4 µg of unpolymerized vimentin. Non-phosphorylated controls were prepared similarly, but without addition of PKA. Following 1 h (21°C) of incubation, non-phosphorylated and PKA-phosphorylated vimentin were spotted onto membranes as above.

Membranes were allowed to dry completely and then blocked 2 h in 5% milk in PBS before overnight incubation with the Alexa-Fluor-680-labeled, phosphorylated or Glu256 and Glu259 MAP2D-pET30. Membranes were washed four times in PBS and bound MAP2D was detected by using the Odyssey Infrared Imaging System (LiCor). Binding data analysis was performed using Microsoft Excel.

Progesterone radioimmunoassay

Progesterone radioimmunoassay (RIA) was performed by the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core, P50HD28934. Results were analyzed by Graphpad Prism^® version 6.01 and are presented as means±s.e.m. on scatter plots.