Sumoylated protein tyrosine phosphatase 1B localizes to the inner nuclear membrane and regulates the tyrosine phosphorylation of emerin

Protein tyrosine phosphatase (PTP)1B is an abundant, non-transmembrane enzyme that plays a major role in regulating insulin and leptin signaling (Bourdeau et al., 2005; Dube and Tremblay, 2005). PTP1B is known to be primarily localized to the outer leaflet of the ER, where it is tethered by a 35-amino-acid hydrophobic C-terminal tail (Frangioni et al., 1992; Woodford-Thomas et al., 1992). Because the ER is an extensive structure spanning the entire cytosol, this localization potentially allows PTP1B access to a wide variety of substrates. For example, PTP1B interacts with N-cadherin at adherens junctions, presumably at sites where these junctions juxtapose the ER (Balsamo et al., 1998; Balsamo et al., 1996; Hernandez et al., 2006; Rhee et al., 2001; Xu et al., 2002). Similarly, PTP1B acts upon proteins in the integrin pathway, such as p130^cas, cortactin and Src (Bjorge et al., 2000; Liu et al., 1996; Mertins et al., 2008). In some cells, PTP1B is cleaved by the protease calpain, which cuts off the hydrophobic tail of PTP1B to release a soluble, more active form of the enzyme (Frangioni et al., 1993; Kuchay et al., 2007).

In the course of studying how PTP1B is regulated, we recently found that PTP1B is inhibited by sumoylation, and that this sumoylation increases upon insulin stimulation (Dadke et al., 2007). We postulated that such growth-factor-induced sumoylation might provide an additional temporal restraint upon PTP1B, effectively prolonging insulin receptor signaling by delaying its dephosphorylation until SUMO proteases restored PTP1B activity through desumoylation. Two large issues, however, remained unresolved. These were: where does PTP1B sumoylation occur, and what intrinsic signals, if any, regulate PTP1B sumoylation? Regarding the first issue, sumoylated proteins are usually found within the nucleus or nuclear envelope, whereas PTP1B is an ER-resident protein. Using fluorescence lifetime imaging microscopy (FLIM), we have previously shown that sumoylated PTP1B accumulates around the nucleus (Dadke et al., 2007). Such localization is not unexpected for an ER-resident protein because the ER is contiguous with the outer nuclear membrane (ONM) of the nuclear envelope.

We report here that, in addition to its known location at the ER surface, PTP1B is present on the inner nuclear membrane (INM), and that this fraction of the total PTP1B population is heavily sumoylated. Furthermore, we show that PTP1B interacts with and regulates the tyrosine phosphorylation of the INM protein emerin. Finally, we also show that during the cell cycle, sumoylation of INM-localized PTP1B increases as cells enter the mitotic phase, as does emerin tyrosine phosphorylation. These results reveal that a portion of PTP1B resides within the nuclear envelope, where, in a sumoylation-dependent manner, it regulates the tyrosine phosphorylation of a key protein that governs nuclear membrane architecture.

Previously, we have shown that PTP1B interacts with PIAS1, a SUMO E3 ligase, thereby catalyzing the sumoylation of PTP1B (Dadke et al., 2007). The presence of a series of slowly migrating bands indicated that PTP1B was sumoylated in multiple sites in cells that coexpressed PTP1B and SUMO-1 (Fig. 1A). The putative sumoylated forms of PTP1B were evident only when the cells were lysed in the presence of N-ethyl-maleimide (NEM), a strong SUMO isopeptidase inhibitor (Dadke et al., 2007). However, the sumoylation “ladder” was markedly reduced when the endogenous enzyme was replaced with a mutant form of PTP1B in which lysine residues within four consensus sumoylation sites (K73, K335, K347 and K389) are mutated to arginine (K4R mutant). Coexpression of wild-type PTB1B (PTP1B-WT) with a non-conjugatable SUMO mutant (SUMO-QT) completely abolished PTP1B sumoylation (Fig. 1A).

To determine whether SUMO modification of PTP1B is related to its catalytic activity, Ptp1b-null mouse embryo fibroblasts (MEFs) were transiently transfected with a catalytically inactive, hemagglutinin (HA)-tagged mutant of PTP1B in which cysteine 215 was replaced by serine (HA-PTP1B-CS) and with T7-SUMO-1, either alone or together. Cell lysates were subjected to anti-PTP1B antibody immunoprecipitation followed by immunoblotting using anti-T7 antibodies to visualize sumoylated PTP1B. PTP1B-WT and PTP1B-CS showed similar levels and patterns of sumoylation (Fig. 1B), indicating that PTP1B sumoylation does not depend on its enzymatic activity. Similar results were obtained in HeLa cells transfected with HA-PTP1B-CS and T7-SUMO-1, followed by anti-HA antibody immunoprecipitation and anti-T7 antibody immunoblotting (supplementary material Fig. S1).

Using FLIM, we previously showed that sumoylated PTP1B is concentrated around the nucleus (Dadke et al., 2007). Because most sumoylated proteins are found within the nucleus or nuclear envelope, whereas PTP1B is attached to the outer leaflet of the ER, we wondered whether sumoylated PTP1B accumulated at the ONM, which is contiguous with the ER. We therefore studied the localization of PTP1B and GFP–SUMO in MEFs. As expected, endogenous PTP1B was found on reticular structures throughout the cytosol, as well as surrounding the nucleus (Fig. 2A). This localization reflects authentic PTP1B distribution, as no staining was seen in Ptp1b^−/− MEFs. GFP–SUMO was seen mostly in the nucleus, but also in the nuclear envelope, where it colocalized with PTP1B in a confocal Z-section image (Fig. 2A and supplementary material Fig. S2).

Next, we stained cells with antibodies against endogenous lamin A/C and PTP1B. Lamin is located in the nuclear lamina, just within the INM, and is therefore not expected to colocalize with ER proteins. To our surprise, we found that lamin A/C and PTP1B also colocalized by immunofluorescence (Fig. 2B). Similar findings were obtained with exogenous wild-type or K4R mutant (sumoylation-resistant) PTP1B (supplementary material Fig. S3), suggesting that the localization of PTP1B to the nuclear envelope is not regulated by sumoylation. These unexpected findings prompted us to examine the subcellular distribution of PTP1B in greater detail. We looked at whether PTP1B had the same cellular distribution as typical ER-resident proteins by staining MEFs with anti-KDEL, anti-PTP1B and anti-lamin antibodies. KDEL and PTP1B colocalized throughout the cytosol in a typical ER pattern, except at the nuclear envelope, where PTP1B staining was much more pronounced than KDEL staining (Fig. 2C). Lamin A/C was exclusively found in a tight band around the nucleus, overlapping the nuclear envelope signal of PTP1B. No overlapping signal was seen between lamin A/C and KDEL.

To supplement these findings, we fractionated purified nuclear envelopes into outer and inner nuclear membranes. Treatment of nuclear envelopes with 1% Triton X-100 is reported to solubilize ER and ONM proteins, leaving the INM and nuclear lamina intact (Dreger et al., 2001; Otto et al., 2001). By contrast, treatment of nuclear envelopes with chaotropic reagents such as urea has the opposite effect, solubilizing the INM and lamina, while leaving the ER and ONM intact (Dreger et al., 2001; Otto et al., 2001). We used this method to assess the location of PTP1B in the nuclear envelope. As expected, a typical ER protein, calnexin, was solubilized by Triton X-100 and not by urea, whereas lamin A/C was solubilized by urea, but not Triton X-100 (Fig. 3A). In contrast to these proteins, PTP1B was not fully solubilized by either treatment. These findings are consistent with the idea that PTP1B resides in the ER and ONM and also in the INM.

Because FLIM measurements suggested that sumoylated PTP1B accumulates in the vicinity of the nuclear envelope (Dadke et al., 2007), we wondered whether this form of PTP1B was distributed throughout the nuclear envelope, or was restricted to the fraction containing ER and ONM (ER/ONM fraction) or to the INM fraction. T7-tagged SUMO-1 was expressed in Ptp1b-null MEFs, which were then fractioned into ER/ONM and INM as described above. Interestingly, all of the sumoylated PTP1B was found in the Triton X-100 insoluble fraction, representing the INM (Fig. 3B). These results indicate that sumoylated PTP1B has a limited intracellular distribution and suggests a function for PTP1B at the INM.

Because we found PTP1B at the INM, and immunofluorescence experiments showed apparent colocalization of PTP1B and lamin A/C, we asked whether PTP1B physically associates with INM proteins. We focused in particular on emerin, a lamin-binding protein that is associated with the X-chromosome-linked Emery-Dreifuss muscular dystrophy (EDMD) (Bione et al., 1994). Several tyrosine phosphorylation sites on emerin have been identified (Schlosser et al., 2006). One tyrosine residue, Tyr95, is of particular interest because small deletions that include this tyrosine residue (Δ95–99) have been found in emerin in human EDMD patients (Ellis et al., 1998). To look at whether PTP1B interacts with emerin, PTP1B was immunoprecipitated from a HeLa cell lysate and probed with anti-emerin antibodies. Using MEFs, we found that emerin co-immunoprecipitates with PTP1B and with a known partner, lamin A/C, but not with control IgG (Fig. 4A). Interestingly, the same degree of binding was observed in pervanadate-treated cells and in cells transfected with a catalytically inactive, substrate-trapping mutant of PTP1B (PTP1B-CS; supplementary material Fig. S4). These results demonstrate that PTP1B and emerin interact and that this interaction requires neither phosphatase activity of PTP1B nor tyrosine phosphorylation of emerin.

To examine whether PTP1B regulates the tyrosine phosphorylation of emerin, we first asked whether recombinant PTP1B could dephosphorylate emerin in vitro. Cells were transfected with a GFP-emerin expression vector or a GFP control vector, then treated with pervanadate, a potent tyrosine phosphatase inhibitor, prior to cell lysis. Total emerin was immunoprecipitated with anti-emerin antibody. Phosphorylated emerin from the anti-emerin antibody immunoprecipitation was incubated with either control GST or GST–PTP1B recombinant protein in an in vitro phosphatase assay. Tyrosine phosphorylation levels of emerin were determined by immunoblot with anti-Tyr-P antibodies. GFP–emerin was tyrosine phosphorylated readily in pervadate-treated cells, and treatment with GST–PTP1B, but not GST, greatly reduced tyrosine phosphorylation of emerin (supplementary material Fig. S5). Next, we examined the levels of phosphotyrosyl emerin in cells expressing exogenous PTPT1B, PTP1B-CS or T7-SUMO-1. Expression of exogenous PTP1B reduced tyrosine phosphorylation of emerin, whereas expression of catalytically inactive PTP1B-CS slightly elevated tyrosine phosphorylation of emerin (Fig. 4B). Interestingly, cells transfected with T7-SUMO-1 also showed increased emerin tyrosine phosphorylation. Given that sumoylation of PTP1B increases under such conditions, and that sumoylation inhibits PTP1B activity (Dadke et al., 2007), these results suggest that PTP1B might dephosphorylate emerin at the INM in a sumoylation-regulated manner.

To further demonstrate the role of PTP1B in tyrosine dephosphorylation of emerin, HeLa cells were stably transfected with a scrambled short hairpin RNA (shRNA) or shRNAs designed specifically to knock down PTP1B. The tyrosine phosphorylation level of emerin in these cells was measured. Scrambled shRNA had no effect on PTP1B protein levels, although one out of the two PTP1B shRNAs significantly reduced the PTP1B expression (Fig. 4C). Lysates from the knockdown cells were subject to Tyr-P immunoprecipitation and anti-emerin antibody immunoblotting. We found that the tyrosine phosphorylation level of emerin was reciprocal to PTP1B expression level; reduced PTP1B correlated with increased tyrosine-phosphorylated emerin (Fig. 4C).

Because sumoylation is known to inhibit PTP1B activity (Dadke et al., 2007), we asked whether sumoylation of PTP1B affects emerin tyrosine phosphorylation. HeLa cells were transfected with HA-PTP1B-WT or HA-PTP1B-K4R alone or in combination with T7-SUMO-1, followed by measurement of the relative tyrosine phosphorylation level of emerin in these cells. As expected, we found that PTP1B-WT, but not the PTP-K4R mutant, was sumoylated when the cells were co-transfected with T7-SUMO-1 (Fig. 5, top panel). To measure the tyrosine phosphorylation level of emerin, cell lysates were subjected to anti-Tyr-P antibody (clone PY20) immunoprecipitation followed by anti-emerin antibody immunoblot. Cells expressing PTP1B-WT plus T7-SUMO showed markedly increased tyrosine phosphorylation of emerin (Fig. 5, second panel, lane 4), whereas cells expressing a non-sumoylatable form of PTP1B (PTP1B-KR) showed much lower levels of emerin tyrosine phosphorylation (Fig. 5, second panel, lane 5), suggesting that PTP1B regulates the tyrosine phosphorylation of emerin in a sumoylation-dependent manner.

We have previously shown that PTP1B sumoylation can be augmented by growth factors such as insulin (Dadke et al., 2007). To determine whether intrinsic signals regulate SUMO modification of PTP1B, we asked whether PTP1B sumoylation levels change during the cell cycle. We transfected Ptp1b-null MEFs with PTP1B-WT and T7-SUMO-1, and synchronized the cell cycle using thymidine treatment, followed by an overnight nocodazole block to achieve G2–M arrest. These mitotically blocked cells were then washed and released into nocodazole-free media for the indicated time, and cell lysates subjected to anti-PTP1B antibody immunoprecipitation followed by anti-T7 antibody immunoblot. We found that PTP1B sumoylation markedly increased as the cells entered mitosis (Fig. 6). Upon release from mitotic block, PTP1B sumoylation levels dropped to a low level within 30 minutes. These results show that PTP1B sumoylation is temporally regulated in a cell-cycle-dependent manner. Interestingly, the tyrosine phosphorylation levels of emerin followed a similar pattern, increasing in mitotic cells and reverting to basal levels as cells exited mitosis into G1 (Fig. 6).

In this paper, we show a new and unexpected localization for PTP1B, with a novel role at the INM. Previously, PTP1B has been shown to be tethered to the outer leaflet of the ER, with the bulk of the protein positioned towards the cytosol (Frangioni et al., 1992; Woodford-Thomas et al., 1992). Because the ER is an extensive structure that spans the entire cytosol, this localization is not necessarily restrictive in terms of substrate access. For example, ER-associated PTP1B has been reported to localize to newly formed matrix adhesions (Hernandez et al., 2006). In addition, PTP1B also exists in a purely soluble form as a result of calpain-mediated cleavage of the C-terminal ER anchor sequence (Frangioni et al., 1993; Kuchay et al., 2007). PTP1B has not, however, previously been reported inside the nucleus or within the INM. Here, we show by both microscopy and biochemical methods that PTP1B is present at the INM, where it is heavily sumoylated in a cell-cycle-dependent manner. Furthermore, we show that PTP1B interacts with and potentially dephosphorylates emerin, a key INM protein that regulates nuclear architecture. PTP1B binds to emerin constitutively, and sumoylation of PTP1B does not affect its association with emerin (supplementary material Fig. S4). Given the effect of sumoylation on PTP1B activity, and the association of this phosphatase with emerin, it is plausible that PTP1B plays a role in regulating the tyrosine phosphorylation levels of this INM protein.

Previously, we had shown that PTP1B is regulated by sumoylation (Dadke et al., 2007). FLIM data showed an accumulation of sumoylated PTP1B around the nucleus, but the resolution of this method was not sufficient to distinguish a perinuclear distribution from regions of the nuclear envelope (Dadke et al., 2007). In this work, we found that sumoylated PTP1B resides predominantly in the INM, either because that is where it becomes sumoylated or because, once sumoylated, PTP1B accumulates at this location. This event does not require phosphatase activity because a catalytically dead mutant form of PTP1B localized normally. The INM-localized PTP1B can be easily overlooked because we found that excess overexpression of PTP1B can mask the nuclear envelope rim staining with mostly ER signaling in any given fluorescence microscopy analysis. Also, this localization was not dependent on sumoylation because a sumoylation-resistant form of PTP1B also localized to the INM (supplementary material Fig. S3). In this respect, PTP1B does not resemble Ran-GAP, which associates with the nuclear envelope only when sumoylated (Mahajan et al., 1998).

Given that PTP1B is inhibited by sumoylation, and that sumoylation of PTP1B increases during mitosis, our results suggest that the INM pool of PTP1B is regulated in a cell-cycle-dependent manner. If so, what might be the function of this sumoylated pool of PTP1B? Certain proteins of the INM, including emerin and lamin A/C, are known to be tyrosine phosphorylated (Otto et al., 2001; Schlosser et al., 2006). Unlike the analogous cell-cycle-dependent serine/threonine phosphorylations (Ellis et al., 1998; Fields and Thompson, 1995), the physiological relevance of these tyrosine phosphorylations are not well understood. However, there are clues that hint at significant functions for these modifications. For example, Tyr95 in emerin has been mapped as a site of tyrosine phosphorylation (Schlosser et al., 2006). This site lies within a small region (amino acids 95–99) that is deleted in some forms of X-chromosome-linked EDMD (Bengtsson and Wilson, 2004; Ellis et al., 1998). This same region is required for stable interaction of emerin with lamin, and loss of this region results in weakened association of emerin with the nuclear envelope (Bengtsson and Wilson, 2004; Lee et al., 2001). In addition, emerin has been shown to be tyrosine phosphorylated at multiple sites by Src and Abl (Schlosser et al., 2006; Tifft et al., 2009), and it has been shown that these phosphorylations significantly hinder emerin binding to barrier-to-autointegration factor (BAF), a protein that is crucial for targeting the nuclear membrane proteins to chromatin at the early stage of nuclear assembly (Hirano et al., 2005; Margalit et al., 2007; Margalit et al., 2005). In this setting, sumoylation might be important to temporally inhibit the local INM pool of PTP1B during the mitotic portion of the cell cycle, in order to maintain emerin tyrosine phosphorylation and interactions with lamins and BAF (Fairley et al., 2002). Upon completion of mitosis, we speculate that the INM pool of PTP1B is activated by desumoylation and therefore able to dephosphorylate emerin, affecting its association with proteins such as lamin A/C and BAF. Although Ptp1b^−/− MEFs do not exhibit notable defects in nuclear structure (data not shown), it should be remembered that even emerin^−/− MEFs display relatively subtle defects (Lammerding et al., 2005). In addition, it is possible that related PTPs, such as T cell (TC)-PTP, share similar functions with PTP1B at the INM.

Because emerin and lamin play important roles in mechanotransduction and in regulating transcription, our findings also raise the interesting possibility that PTP1B has previously unappreciated nuclear functions and that these functions are regulated by sumoylation. A comprehensive analysis of nuclear protein tyrosine phosphorylation in wild-type versus Ptp1b-null cells might reveal the identity of new targets for this enzyme.

Rabbit polyclonal antibodies against lamin A/C, emerin and HA-probe (Y-11) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal anti-Tyr-P (clone 4G10) and anti-T7 tag antibodies were purchased from Millipore (Wemecula, CA) and Novagen (Madison, WI), respectively. ER membrane marker, Calnexin antibody, was commercially available from Abcam (Cambridge, MA). Rabbit polyclonal antibody against human PTP1B was purchased from R&D System (Minneapolis, MN). Secondary antibody conjugated to peroxidase was purchased from Jackson ImmunoResearch Laboratory (West Grove, PA).

HeLa cells were obtained from the ATCC. Ptp1b^+/+ and Ptp1b^−/− MEFs were a gift from Ben Neel (Haj et al., 2003). These cells were maintained in Dulbecco's modified Eagle's medium (Gibco BRL) supplemented with 10% fetal bovine serum, 50 U/ml penicillin, 50 μg/ml streptomycin and 200 μg/ml hygromycin B.

Expression vectors encoding T7-tagged SUMO1, wild-type pCMV6-HA-PTP1B and the sumoylation-resistant mutant HA-PTP1B-K4R have been described previously (Dadke et al., 2007). pEGFP–emerin was a gift from Yuichi Tsuchiya and Kiichi Arahata (National Institute of Neuroscience, Tokyo, Japan). pEGFP–lamin A was a gift from Susan Michaelis (Johns Hopkins University School of Medicine, Baltimore, MD). Transfections of HeLa cells and MEFs were carried out using Lipofectamine 2000, according to the manufacturer's specifications.

For immunofluorescence, the transfected cells on coverslips were fixed in cold methanol (−20°C) for 10 minutes at 24 hours after transfection. Cell membranes were then permeabilized by incubating with 0.5% NP40 in phosphate-buffered saline (PBS) 10 minutes prior to immunostaining. Confocal microscopy was performed using a Nikon Eclipse TE2000 inverted microscope with confocal laser optics. The images were captured using a 60× oil objective (Nikon, 60× NA 1.4).

Cells were lysed in RIPA buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 50 mM HEPES, 10% glycerol, 1% Triton X-100, 0.1% SDS, 50 mM NaF, 10 mM β-glycerol phosphate, 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride and 10 μg/ml aprotinin). Lysates were clarified by centrifugation at 25,000 g for 2 minutes at 4°C. Protein concentrations were determined using a bicinchoninic acid protein assay kit (Pierce Chemical). For immunoprecipitations, lysates were incubated with the appropriate antibodies overnight at 4°C. Immune complexes were collected onto Protein A-Sepharose beads, washed extensively, resolved by SDS-PAGE and transferred onto Immobilon-P membranes (Millipore, Bedford, MA). Immunoblots were blocked with 3% bovine serum albumin or 5% Carnation non-fat dry milk in 10 mM Tris-HCl pH 7.4, 150 mM NaCl and 0.05% Tween 20. After incubation with appropriate primary and secondary antibodies, blots were visualized using enhanced chemiluminescence reagents (ECL; Amersham Biosciences). Quantification was carried out using NIH ImageJ software, version 1.40. Data are expressed as relative units of phosphorylated protein per total protein for each band. Anti-PTP1B and anti-HA antisera were used at 1:1000 for immunoblotting. Anti-human PTP1B antibodies were used at 1 μg/ml for immunoprecipitations. All other antibodies were used at concentrations recommended by the supplier.

Nuclei and nuclear envelopes were prepared as previously described (Emig et al., 1995; Otto et al., 1992). Briefly, cells were washed and scraped in cold PBS, centrifuged, and the cell pellet suspended in sucrose containing STM 0.25 (50 m Tris-HCl pH 7.4, 0.25 M sucrose, 5 mM MgSO₂ and 2 mM DTT) plus protease inhibitors and NP40, followed by Dounce homogenization. The homogenate was adjusted to 1.4 M sucrose and overlaid between STM 2.1 and STM 0.8 cushions, and filled up with STM 0.25 buffer. Nuclear envelopes were obtained by centrifugation at 100,000 g for 1 hour. The sedimented material was then treated with 250 μg/ml heparin and 400 U DNAase I for 90 minutes at 4°C. Nuclear envelopes were further subfractionated using a published procedure, identifying tyrosine phosphorylated proteins in the INM and ONM/ER fractions (Dreger et al., 2001). The samples were extracted either with 0.5% (wt/vol) Triton X-100 or with urea and Na₂CO₃ (final concentrations of 4 M and 0.1 M, respectively) for 15 minutes at 4°C. Triton X-100-resistant material was pelleted at 22,000 g whereas chaotrope-resistant material was sedimented at 100,000 g.

Ptp1b^−/− MEFs co-transfected with PTP1B-WT and T7-SUMO-1 were synchronized to G1–S phase by treating the culture with 2 mM thymidine (final concentration) at 37°C for 19 hours. To maximize the number of synchronized cells, the cells were washed three times in PBS and the block was released in complete media without thymidine for 8 hours. For G1–S blocked cells, a final concentration of 2 mM thymidine was added to the cell culture again for another 16 hours. For mitotic blocked cells, a final concentration of 100 ng/ml nocodazole was added to the media for 12 hours at 37°C. The rounded, loosely attached mitotic cells were shaken off and collected. Both G1–S and mitotic blocked cells were washed in PBS and released for the indicated times, and then lysed for either anti-PTP1B or anti-Tyr-P (clone 4G10) antibody immunoprecipitations.

We thank Kathy Wilson (Johns Hopkins University School of Medicine, Baltimore, MD) and Ben Neel (Princess Margaret Hospital, Toronto, Canada) for their gifts of emerin cDNAs and wild-type and Ptp1b^−/− MEFs, respectively.