Role of ubiquitylation and USP8-dependent deubiquitylation in the endocytosis and lysosomal targeting of plasma membrane KCa3.1

Abstract

We recently demonstrated that plasma membrane KCa3.1 is rapidly endocytosed and targeted for lysosomal degradation via a Rab7- and ESCRT-dependent pathway. Herein, we assess the role of ubiquitylation in this process. Using a biotin ligase acceptor peptide (BLAP)-tagged KCa3.1, in combination with tandem ubiquitin binding entities (TUBEs), we demonstrate that KCa3.1 is polyubiquitylated following endocytosis. Hypertonic sucrose inhibited KCa3.1 endocytosis and resulted in a significant decrease in channel ubiquitylation. Inhibition of the ubiquitin-activating enzyme (E1) with UBEI-41 resulted in reduced KCa3.1 ubiquitylation and internalization. The general deubiquitylase (DUB) inhibitor, PR-619 attenuated KCa3.1 degradation, indicative of deubiquitylation being required for lysosomal delivery. Using the DUB Chip, a protein microarray containing 35 DUBs, we demonstrate a time-dependent association between KCa3.1 and USP8 following endocytosis, which was confirmed by coimmunoprecipitation. Further, overexpression of wild-type USP8 accelerates channel deubiquitylation, while either a catalytically inactive mutant USP8 or siRNA-mediated knockdown of USP8 enhanced accumulation of ubiquitylated KCa3.1, thereby inhibiting channel degradation. In summary, by combining BLAP-tagged KCa3.1 with TUBEs and DUB Chip methodologies, we demonstrate that polyubiquitylation mediates the targeting of membrane KCa3.1 to the lysosomes and also that USP8 regulates the rate of KCa3.1 degradation by deubiquitylating KCa3.1 prior to lysosomal delivery.—Balut, C. M., Loch, C. M., Devor, D. C. Role of ubiquitylation and USP8-dependent deubiquitylation in the endocytosis and lysosomal targeting of plasma membrane KCa3.1.

The intermediate-conductance, Ca²⁺-activated K⁺ channel (KCa3.1) plays a critical role in the functional activity of cells from a wide variety of tissues and disease states, where it modulates the calcium signaling cascade by regulating membrane potential. As such, there is intense interest in understanding the function and regulation of KCa3.1, with the goal of developing therapeutic strategies in various disease states (1).

One factor in determining the physiological response of an ion channel is the number of active channels present on the cell surface at any given time. Thus, trafficking pathways, endocytosis, recycling, and degradation all work in concert to maintain appropriate channel numbers. Although considerable progress has been made in understanding the mechanisms of assembly and anterograde trafficking of KCa3.1 (2, 3), the details of KCa3.1 postendocytic trafficking and sorting have just begun to be unraveled. Recently, using a novel biotin ligase acceptor peptide (BLAP)-tagged KCa3.1, which allows the endocytosis of the channel to be followed in real time, we have shown that KCa3.1 is rapidly internalized from the plasma membrane and targeted for lysosomal degradation via a Rab7- and multivesicular body (MVB)/endosomal sorting complex required for transport (ESCRT)-dependent pathway (4). However, to date, nothing is known about the sorting signal that engages KCa3.1 into the MVB degradative pathway.

Covalent modification by ubiquitin has been established as a sorting signal that targets many membrane proteins to the lysosomes through the MVB (5). The ubiquitylation of such cargo often begins at the plasma membrane and continues on endosomal membranes (6). The ubiquitylation of some membrane proteins may actually serve as an internalization signal (7), although other ways of triggering the endocytosis of specific proteins have been reported (8). In the current model of ubiquitin-directed lysosomal degradation of membrane proteins, the ubiquitylated cargo is recognized and sorted on the endosomal membrane by members of the ESCRT machinery, which direct the cargo into the invaginating MVB vesicles (9). However, while ubiquitylation serves as a predominant sorting signal for many receptors and membrane proteins, some cargo does not require ubiquitylation to enter the ESCRT-dependent pathway (10–12), emphasizing the complexity of the MVB sorting process.

On the basis of this, we have undertaken studies to elucidate the sorting signal involved in endocytosis and lysosomal targeting of KCa3.1. Using a novel method of isolating ubiquitylated proteins from cell lysates, based on tandem ubiquitin binding entities (TUBEs; ref. 13), in combination with the BLAP-KCa3.1 methodology (4, 14), we demonstrate that ubiquitylation serves as a signal for KCa3.1 endocytosis and that the channel becomes polyubiquitylated following endocytosis. Furthermore, by using a novel protein microarray, the DUB Chip, we demonstrate that endocytosed KCa3.1 is targeted for USP8-mediated deubiquitylation before delivery to lysosomes. The present study delineates, for the first time, the role of ubiquitylation and USP8-dependent deubiquitylation in regulating the postendocytic trafficking and lysosomal degradation of KCa3.1.

MATERIALS AND METHODS

Molecular biology

Insertion of the BLAP sequence (GLNDIFEAQKIEWHE) into the extracellular loop between transmembrane domains S3 and S4 of KCa3.1 has been previously described (14). The C-terminal myc epitope-tagged KCa3.1 was previously described (3). The GFP-tagged USP8 constructs [wild type (WT) and catalytically inactive, C786S] were generously provided by Dr. S. Urbé (University of Liverpool, Liverpool, UK). The hemagglutinin (HA)-tagged WT ubiquitin construct (15) was obtained from Addgene (plasmid 17608; Addgene, Cambridge, MA, USA).

Cell culture

Human embryonic kidney (HEK293) cells were obtained from the American Type Culture Collection (Manassas, VA, USA). The cells were cultured and transfected as described previously (14).

Antibodies

Polyclonal α-streptavidin Ab was obtained from Genscript Corporation (Piscataway, NJ, USA). Monoclonal HA (HA.11) and c-myc (clone 9E10) antibodies were obtained from Covance (Richmond, CA, USA). Monoclonal α-tubulin and polyclonal α-USP8 were obtained from Sigma-Aldrich (St. Louis, MO, USA). Monoclonal α-GFP was obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Monoclonal antibody against ubiquitinylated proteins (clone FK2) was obtained from Millipore (Billerica, MA, USA).

Biotinylation of KCa3.1 using BirA

BLAP-tagged KCa3.1, heterologously expressed in HEK293 cells, was enzymatically biotinylated using recombinant biotin ligase (BirA), as described previously (14). BirA was expressed from pET21a-BirA (generously provided by Dr. Alice Y. Ting, Massachusetts Institute of Technology, Cambridge, MA, USA) in Escherichia coli, according to previously published methods (16). Plasma membrane BLAP-tagged KCa3.1 was then labeled with streptavidin-Alexa488 or streptavidin-Alexa555 (Invitrogen, Carlsbad, CA, USA), and the cells were either incubated for various periods of time at 37°C, as indicated in the text, or immediately fixed and permeabilized (14). Nuclei were labeled with DAPI (Sigma-Aldrich). Cells were subjected to laser confocal microscopy using an Olympus FluoView 1000 system. To ensure maximal X-Y spatial resolution, sections were scanned at 1024 × 1024 pixels, with sequential 3-color image collection to minimize crosstalk between the channels imaged.

Immunoprecipitations (IPs) and immunoblots (IBs)

Our IP and IB protocols have been previously described (2, 14, 17, 18). Briefly, cells were lysed, and equivalent amounts of total protein were precleared with protein G-agarose beads (Invitrogen) and incubated with the indicated antibody. A nonspecific IgG was used as a negative control. Immune complexes were precipitated with protein G-agarose beads, and the proteins were resolved by SDS-PAGE followed by IB. To eliminate interference by the heavy and light chains of the immunoprecipitating antibody in the IP, mouse IgG Trueblot Ultra (eBioscience, San Diego, CA, USA) was used as a secondary antibody for the detection of immunoprecipitated proteins in the IB.

Determination of degradation rate for plasma membrane-localized KCa3.1

The degradation rate for endocytosed membrane KCa3.1 was determined as described previously (14). Briefly, the channel was specifically biotinylated using BirA and labeled with streptavidin, as above. Next, cells were incubated for various periods of time at 37°C, as indicated. The cells were then lysed, and equivalent amounts of total protein were separated by SDS-PAGE, followed by IB for streptavidin. As streptavidin remains tightly coupled to the channel during SDS-PAGE, this provides a direct correlate to the IF studies detailed above. Bands were quantified by the densitometry function of Quantity One software (Bio-Rad, Hercules, CA, USA). Data were corrected for background and nonspecific labeling (measured on cells treated only with streptavidin, in the absence of biotinylation). The blots were also probed and corrected for tubulin, as a protein loading control. The obtained band intensities for the various time points were normalized relative to the intensity at time 0 and reported.

Pulldown of ubiquitylated KCa3.1 using TUBEs

To determine the ubiquitylation state of KCa3.1 at the plasma membrane, and following endocytosis, we used GST-tagged TUBEs (LifeSensors Inc., Malvern, PA, USA). TUBEs display up to a 1000-fold increase in affinity for polyubiquitin moieties over the single ubiquitin binding-associated domain (UBA) and can be used to both stabilize polyubiquitin chains and efficiently purify ubiquitylated proteins from cell lysates (13). For these studies, the channel was enzymatically biotinylated and streptavidin labeled at the plasma membrane, as described above, after which the cells were immediately lysed in the presence of GST-TUBE2 (200 μg/ml), or returned to 37°C for various periods of time to allow endocytosis to occur, and then lysed in the presence of TUBEs. The TUBEs were subsequently pulled down on glutathione agarose beads, by incubating equivalent amounts of total protein with the beads for 2 h at 4°C on a rocking platform. The pulldown was then subjected to SDS-PAGE, and the resulting IB was probed with α-streptavidin Ab. In this way, only the streptavidin-tagged KCa3.1, which was ubiquitylated and hence bound to TUBEs, was detected.

Detection of the DUBs interacting with KCa3.1 by using the DUB Chip microarray

The DUB Chip (LifeSensors) is a protein microarray printed in triplicate, at two separate concentrations, with 35 DUBs, as well as several deSUMOylases and deNeddylases. The details of the microarray production were previously described (19). To define the specific DUBs involved in deubiquitylating endocytosed KCa3.1, plasma membrane BLAP-tagged KCa3.1 was labeled with streptavidin-Alexa488, as described above, and the cells were either immediately lysed in the presence of GST-TUBE2 (200 μg/ml), or returned to 37°C for various periods of time to allow endocytosis to occur, and then lysed in the presence of TUBEs. Cells labeled only with streptavidin in the absence of enzymatic biotinylation were used as a control for nonspecific labeling. The TUBEs, with ubiquitylated cargo attached, were pulled down on glutathione agarose, eluted (50 mM Tris 8.0, 10 mM reduced glutathione), and then directly hybridized to the DUB Chip, followed by thorough washes (PBST, PBS, ddH₂O ×2). Given that the only fluorescent protein in the sample is KCa3.1, an interaction between KCa3.1 and a specific DUB on the microarray was evaluated for each time point by scanning the arrays on a Typhoon 9410 Imager (GE Life Sciences, Piscataway, NJ, USA) with 488-nm excitation and 526-nm emission at 10-μm resolution. For array analysis, the image files of scanned arrays were digitally quantified, and data from two independent arrays under each experimental condition were combined for analysis. Signal intensity was then calculated as feature intensity minus local background (measured on control cells), quantity divided by local background [(F-B)/B]. Signals were then normalized to the median background signal of all features present on the subarray. The normalized data from 6 printed spots of each DUB/dilution combination were then summarized by their median, and the standard deviation was calculated. These data were then rendered graphically.

Short-interfering RNA (siRNA) treatment

The siGENOME smartpool siRNA (a mixture of 4 oligonucleotide duplexes) against the coding region of human USP8 was obtained from Dharmacon Research (Chicago, IL, USA). The siGENOME Non-Targeting siRNA Pool no. 2 (Dharmacon Research) was used as control. HEK293 cells, stably expressing BLAP-KCa3.1, were plated at ∼50% confluence and transfected with 50 nM siRNA duplexes using DharmaFECT 1, according to the manufacturer's directions. Experiments were carried out at 72 h post-transfection.

Drug treatments

To inhibit KCa3.1 endocytosis, cells were preincubated with hypertonic sucrose (0.45 M) for 15 min at 4°C and then returned for 90 min at 37°C in the presence of sucrose. To inhibit the ubiquitin-activating enzyme E1, cells were preincubated with 50 μM 4[4-(5-nitro-furan-2-ylmethylene)-3,4-dioxo-pyrazolidin-1-yl]-benzoic acid ethyl ester (UBEI-41; Biogenova, Rockville, MD, USA) for 60 min at 37°C. Next, KCa3.1 was enzymatically biotinylated at the cell surface and streptavidin labeled, and cells were returned for 90 min at 37°C in the presence of UBEI-41. To block DUB activity, cells were treated for the indicated periods of time at 37°C with 50 μM PR-619 (LifeSensors), a cell-permeable, pan-DUB inhibitor (20).

Chemicals

All chemicals were obtained from Sigma-Aldrich, unless otherwise stated.

Statistics

All data are presented as means ± se, where n indicates the number of experiments. For DUB Chip data analysis, a standard unpaired, 1-sided Student's t test (10 degrees of freedom) was used to evaluate signal between each time point and the control (cells treated with streptavidin in the absence of enzymatic biotinylation), where the null hypothesis was [signal control]=[signal at time X]. To compare the normalized values of the IB band intensities, statistical analysis was performed using the nonparametric Kruskal-Wallis test. Values of P < 0.05 were considered statistically significant and are reported.

RESULTS

KCa3.1 is ubiquitylated during endocytic trafficking

As recently described (4, 14), we followed the endocytic trafficking of BLAP-KCa3.1 by enzymatically biotinylating the channel at the cell surface using recombinant BirA and labeled with streptavidin-Alexa 555. As shown in Fig. 1A, at time 0 min, KCa3.1 is exclusively labeled at the plasma membrane and is almost completely endocytosed after 90 min incubation at 37°C.

Figure 1.

Membrane KCa3.1 is ubiquitylated during endocytic trafficking. A) At time 0 min, KCa3.1 is localized at the plasma membrane (red) and is completely endocytosed after 90 min incubation at 37°C. B) To detect ubiquitylated KCa3.1, plasma membrane channel was enzymatically biotinylated and labeled with streptavidin; cells at time 0 and 90 min were lysed in the presence of GST-TUBE2. Following pulldown, the immunoblot (IB) was probed using α-streptavidin Ab to identify BLAP-tagged KCa3.1/streptavidin. To control for nonspecific labeling, KCa3.1- expressing cells were labeled with streptavidin in the absence of enzymatic biotinylation (control lanes in all IB). No channel is detected in these controls, confirming the specificity of our labeling protocol. As shown in lane 3, endocytosed KCa3.1 is heavily ubiquitylated (time 90 min) compared with channel present at the plasma membrane (time 0 min; lane 2). Importantly, similar levels of KCa3.1 were detected for the two time points, as assessed by IB of total cell lysate (lanes 5 and 6). C) Samples were prepared and lysed as in B, then KCa3.1 was immunoprecipitated using streptavidin Ab, and the subsequent IB was probed using α-ubiquitin Ab. KCa3.1 is strongly ubiquitylated following endocytosis. D) HEK cells were doubly transfected with BLAP-KCa3.1 and HA-ubiquitin. Ubiquitin was immunoprecipitated using an anti-HA Ab and subsequently IB for streptavidin-tagged KCa3.1. E) KCa3.1 was immunoprecipitated using an anti-streptavidin antibody and the subsequent IB probed with anti-HA Ab.

To rapidly assess the ubiquitylation status of KCa3.1 at the plasma membrane and following endocytosis, we have combined the capabilities of the BLAP-KCa3.1 construct with the advantages of TUBEs, which enable fast isolation of ubiquitylated proteins from cell lysates (13). KCa3.1 was biotinylated and labeled with streptavidin and allowed to internalize for 90 min at 37°C, as above. The cells were then lysed in the presence of GST-TUBE2, and the TUBEs were pulled down on glutathione-agarose beads, followed by IB using α-streptavidin Ab. This approach allows the identification of the KCa3.1-streptavidin complex only if the channel was ubiquitylated and, hence, bound to TUBEs. To control for nonspecific labeling, KCa3.1-expressing cells were labeled with streptavidin, in the absence of enzymatic biotinylation. As shown in Fig. 1B (lane 1), we did not detect any nonspecific band during IB, confirming the specificity of our labeling protocol. At time 0 min, when the labeled channel is exclusively localized at the cell surface, we observed 2 faint bands (Fig. 1B, lane 2). The first band (∼90 kDa) corresponds to the molecular mass for streptavidin-tagged KCa3.1, while the second band, double in size, most likely represents the channel dimmer, as previously reported for other K⁺ channels, including KCa2.3, another member of the Ca²⁺-activated K⁺ channel family (21, 22). The low amount of protein detected in these lanes is indicative of a low amount of ubiquitylated KCa3.1 in the plasma membrane. However, following endocytosis for 90 min, we observed a significant increase in the amount of protein detected (Fig. 1B, lane 3), indicating that KCa3.1 was heavily ubiquitylated following endocytosis. Note that similar amounts of membrane KCa3.1 are detected at time 0 and 90 min, as assessed by IB of the total cell lysate with α-streptavidin Ab (Fig. 1B, lanes 5 and 6), indicating that the difference in ubiquitylation observed is not due to a difference in the plasma membrane channel expression. Results shown are representative of >10 experiments.

To confirm ubiquitylation of KCa3.1 following endocytosis, BLAP-KCa3.1 was immunoprecipitated using α-streptavidin antibody and probed using α-ubiquitin antibody. As before, we failed to detect any nonspecific labeling in the absence of enzymatic biotinylation (Fig. 1C, lane 1). Moreover, similar to data in Fig. 1B, we observed a low amount of ubiquitylated channel at time 0 min (Fig. 1C, lane 2), followed by a strong increase in channel ubiquitylation, apparent as a high molecular mass smear, at time 90 min (Fig. 1C, lane 3). Similar results were observed in 3 experiments. Although the results in Fig. 1B, C are qualitatively similar in demonstrating an increased ubiquitylation of the channel following endocytosis, the appearance of the bands is different between the two. That is, we detected discrete bands when we pulled down TUBEs and probed for the channel, as opposed to a more classic smear when we immunoprecipitated the channel and probed for ubiquitin.

To rule out the possibility that the difference observed above was a TUBE-induced artifact, we cotransfected cells with BLAP-KCa3.1 and HA-tagged ubiquitin and performed a classic co-IP in both directions in the absence of TUBEs. To minimize deubiquitylation of KCa3.1 during cell lysis and IP, lysis buffer included 50 μM PR-619, a general DUB inhibitor (20). First, we did an IP for HA-ubiquitin and an IB for streptavidin-tagged KCa3.1. As shown in Fig. 1D, at time 0 min, we observed a faint band consistent with a small amount of ubiquitylated KCa3.1 in the plasma membrane (lane 1), and this was significantly increased following endocytosis for 90 min (lane 2). When we performed the co-IP in the opposite direction, immunoprecipitating KCa3.1 with α-streptavidin Ab and blotting for ubiquitin with α-HA Ab (Fig. 1E), we observed again a heavily ubiquitylated KCa3.1 at time 90 min (Fig. 1E, lane 2). The appearance of the bands in these experiments mirrors the results obtained in the presence of TUBEs (Fig. 1B, C). Taken together, these data clearly indicate that KCa3.1 is heavily ubiquitylated following endocytosis. Moreover, these results demonstrate the efficiency of the TUBE assay in defining the ubiquitylation status of a membrane protein in a more specific and less time-consuming manner, as an immunoprecipitation step is not required.

Ubiquitylation is required for KCa3.1 endocytosis

Our results above demonstrate that KCa3.1 is ubiquitylated during the endocytic process. To determine whether KCa3.1 is specifically ubiquitylated at the plasma membrane, and this acts as a trigger for endocytosis we blocked channel internalization using hypertonic sucrose (23) and analyzed KCa3.1 endocytosis and ubiquitylation as above. As a control, we used cells labeled for the channel at the plasma membrane, in the absence of any sucrose treatment. As shown in Fig. 2A, at time 0 min, the channel was expressed at the cell surface. Within 90 min, the channel was almost completely endocytosed in control cells, yet failed to internalize in cells treated with hypertonic sucrose. As shown in Fig. 2B, hypertonic sucrose also significantly decreased the level of KCa3.1 ubiquitylation after 90 min (top blot, lane 5) compared to control (top blot, lane 3). Similar results were observed in 3 experiments. These data suggest that KCa3.1 may require ubiquitylation as a signal for endocytosis. To confirm this, we used UBEI-41, a cell-permeable inhibitor of the E1-activating enzyme. Cells stably expressing KCa3.1 were pretreated with UBEI-41 (see Materials and Methods), and the channel was labeled at the cell surface as above. As previously reported (24), UBEI-41 dramatically reduced the internalization of KCa3.1 (Fig. 2A), and this correlated with a reduced level of ubiquitylation (Fig. 2B; top blot, lane 4). Similar results were observed in 3 experiments. The effect of UBEI-41 suggests that, as with other membrane proteins, internalization of KCa3.1 may require ubiquitylation of either the channel itself or adaptors as a signal for the subsequent endocytosis.

Figure 2.

Ubiquitylation is required for KCa3.1 endocytosis. A) BLAP-KCa3.1 was enzymatically biotinylated at the cell surface and labeled with streptavidin-Alexa555 or streptavidin and incubated at 37°C for 90 min in the absence or presence of hypertonic sucrose, an inhibitor of endocytosis, or the ubiquitin-activating enzyme (E1) inhibitor UBEI-41. Both hypertonic sucrose and UBEI-41 dramatically reduced the internalization of KCa3.1. B) GST-TUBE pulldown assay, followed by IB using α-streptavidin Ab, suggests that KCa3.1 may be ubiquitylated at the plasma membrane in the absence of endocytosis (compare lanes 5 and 2) and that inhibition of ubiquitylation blocks endocytosis of the channel (UBEI-41 treatment). Subsequent to endocytosis, KCa3.1 becomes heavily polyubiquitinated (T=90 min, lane 3).

KCa3.1 is deubiquitylated prior to lysosomal degradation

The ubiquitylation of KCa3.1 may serve as a sorting signal that targets KCa3.1 to the lumen of lysosomes via MVBs (25). However, many ubiquitylated membrane proteins and receptors undergo deubiquitylation prior to delivery to lysosomes as a way of recycling and preserving the cellular ubiquitin pool (26, 27). To assess the role of DUBs in KCa3.1 degradation, the degradation rate for plasma membrane-localized KCa3.1 was evaluated in the absence or presence of the general DUB inhibitor PR-619 (20). As shown in Fig. 3A, subsequent to endocytosis, membrane KCa3.1 is extensively degraded within 8 h in control cells. That is, only 46 ± 9% (n=3) of the membrane protein was still detected at the end of the incubation period. In contrast, in PR-619 treated cells, channel degradation was significantly inhibited, with 86 ± 6% (Fig. 3A, bar graph; n=3; P<0.05) of membrane KCa3.1 still present. Moreover, PR-619 treatment resulted in an accumulation of ubiquitylated KCa3.1 as shown by the result of the GST-TUBEs pulldown assay (Fig. 3B). These results implicate DUBs as an active component required for proper lysosomal degradation of endocytosed KCa3.1.

Figure 3.

KCa3.1 is deubiquitylated before delivery to lysosomes. A) BLAP-KCa3.1 was enzymatically biotinylated and streptavidin labeled; the channel was allowed to internalize for the times indicated in the absence or presence of PR-619. PR-619 treatment significantly inhibited KCa3.1 degradation (bar graph, n=3). *P < 0.05. B) Cells were treated and labeled as in A, and the ubiquitylation of KCa3.1 was assessed by GST-TUBE pulldown assay. A representative blot is shown, indicating that PR-619 treatment prevents channel deubiquitylation. These data demonstrate that KCa3.1 needs to be deubiquitylated for proper lysosomal degradation.

DUB Chip as a tool to identify specific DUBs interacting with KCa3.1

To identify the specific DUBs involved in deubiquitylating KCa3.1, we combined our BLAP-tagged methodology of labeling the channel at the cell surface with the newly developed DUB protein array (DUB Chip), containing 35 DUBs, as well as several deSUMOylases and deNeddylases (19). KCa3.1 was enzymatically biotinylated at the plasma membrane and labeled with streptavidin-Alexa 488, as above. Cells were lysed in the presence of GST-TUBE2 immediately (time 0 min), or at time points of 90 min and 3 h to allow the channel to be internalized prior to being lysed in the presence of TUBEs. As before, cells expressing the channel and treated with streptavidin-Alexa 488 in the absence of enzymatic biotinylation were used as a control and resulted in no nonspecific signal (Fig. 4A, left panel). Subsequently, the lysates were pulled down on glutathione agarose, eluted, then hybridized on the DUB Chip. Given that the only fluorescent protein in the sample was KCa3.1, the association of the channel with a DUB on the microarray was directly visualized using a Typhoon imaging system (see Materials and Methods). The results of DUB Chip hybridization are shown in Fig. 4B (top panel), where the association between the channel and specific DUBs is observed as a dark spot on the microarray. At time 0 min, we did not detect an interaction between KCa3.1 and any of the DUBs present on the microarray, as expected given the low level of channel ubiquitylation. However, following endocytosis for 90 min, we detected a strong association of KCa3.1 with USP2 and USP8, and a weaker association with AMSH. Moreover, this association was transient in nature, as we failed to observe any signal above background for the samples collected after 3h incubation. Figure 4B (bottom panel) presents the quantification of the DUB Chip data. These results demonstrate, for the first time, an interaction between an ubiquitylated substrate from a cell lysate (KCa3.1) and enzymes (DUBs) on a microarray.

Figure 4.

DUB Chip as a tool to identify specific DUBs interacting with KCa3.1. A) BLAP-KCa3.1 was labeled with streptavidin-Alexa488 and incubated for the times indicated at 37°C. B) Cells were lysed in the presence of GST-TUBE2; the lysates were pulled down on GST beads, eluted, and hybridized on a DUB Chip (top panel). An interaction between fluorescently-tagged KCa3.1 and specific DUBs was quantified by measuring the fluorescence intensity (see Materials and Methods). DUB Chip data, expressed as relative fluorescence units (RFU), indicates an interaction between ubiquitylated KCa3.1 and both USP2 and USP8, and a weaker association with AMSH (bottom panel). C) Co-IP confirmed the interaction between KCa3.1 and USP8. Cells were cotransfected with myc-tagged KCa3.1 and either the WT or DN form of the GFP-USP8. USP8 was immunoprecipitated using either an anti-GFP Ab (lanes 1–3 and 5–6) or a nonspecific IgG (lanes 4 and 7) and subsequently IB for KCa3.1 with α-myc Ab. Total cell lysates are shown in lanes 8 and 9.

Given the well-documented role of USP8 as a DUB involved in deubiquitylating cargo destined for lysosomal degradation (28), we confirmed a role for this DUB on KCa3.1 targeting to the lysosome. We first confirmed an association between KCa3.1 and USP8 by co-IP. HEK cells were cotransfected with myc-tagged KCa3.1 and either the WT or the catalytically inactive (C786S) form of GFP-tagged USP8, which acts as a dominant negative (DN). As shown in Fig. 4C, we observe an association of myc-KCa3.1 with both WT (lane 3) and DN (lane 6) forms of USP8. Similar results were observed in 3 experiments; confirming the association between KCa3.1 and USP8 detected on the DUB array.

USP8 overexpression alters the ubiquitylation and degradation rate of KCa3.1

To further define the role of USP8 in lysosomal degradation of KCa3.1, HEK cells were transfected with BLAP-KCa3.1 and either GFP alone, GFP-USP8 WT, or GFP-USP8 DN. Plasma membrane-localized channel was labeled with streptavidin-Alexa 555 and allowed to internalize for 8 h, after which the cells were examined by confocal microscopy. In agreement with our previous reports (4, 14), KCa3.1 was mostly degraded in control cells after 8 h incubation, as assessed by the reduced fluorescent signal associated with the channel (Fig. 5A, top panel). WT-USP8 exhibited a diffuse, widespread cytosolic signal (Fig. 5A, middle panel), whereas DN-USP8 was localized in enlarged vesicles, as well as the cytosol (Fig. 5A, bottom panel), as previously reported (6, 29). Interestingly, both WT and DN USP8 had a strong inhibitory effect on membrane KCa3.1 degradation, as seen by the clear fluorescent signal associated with the channel after 8 h (Fig. 5A, middle and bottom panels). Moreover, endocytosed KCa3.1 accumulates in the DN-USP8 induced structures, as indicated by the clear colocalization (Fig. 5A, arrow in bottom overlay).

Figure 5.

USP8 overexpression alters the ubiquitylation and degradation rate of KCa3.1 A) Cells were doubly transfected with BLAP-KCa3.1 and either GFP vector alone or the WT or DN GFP-USP8. KCa3.1 was labeled at the cell surface with streptavidin-Alexa555, and cells were incubated at 37°C for 8 h. In cells overexpressing either form of USP8, KCa3.1 degradation is slowed compared with control cells, as shown by the strong intracellular signal. Moreover, KCa3.1 clearly colocalizes with DN GFP-USP8 subsequent to endocytosis (bottom panel, arrow in overlay). B) Degradation rate of membrane KCa3.1 was quantified as above. Both WT and DN GFP-USP8 caused a significant delay in KCa3.1 degradation rate (bar graph, n=3). *P < 0.05. C) To correlate the degradation in B with the level of KCa3.1 ubiquitylation, cells were prepared as in B and lysed in the presence of GST-TUBE2 at the indicated times, followed by pulldown on GST beads and IB using α-streptavidin Ab. WT USP8 overexpression decreases ubiquitylation of KCa3.1, while DN USP8 prevents deubiquitylation, as compared with control cells.

The rate of KCa3.1 degradation in the presence of either GFP, WT, or DN USP8 was quantified. As described previously (4), KCa3.1 was markedly degraded after 8 h in cells expressing GFP (35±6% membrane KCa3.1 still present, n=3; Fig. 5B, top blot). In cells expressing WT USP8 (Fig. 5B, middle blot), the degradation of KCa3.1 was significantly reduced (63±9% membrane KCa3.1 still present, n=3), as well as in the DN USP8-expressing cells (83±11%, n=3, P<0.05; Fig. 5B, bottom blot). These results were then correlated with the level of channel ubiquitylation. Results indicated that the ubiquitylation level of KCa3.1 was strongly decreased at any time point by overexpressing WT USP8 (Fig. 5C, middle blot), and markedly increased by overexpressing its catalytically inactive mutant (Fig. 5C, right blot), as compared with control cells (Fig. 5C, left blot). Similar results were observed in 3 experiments. These data are consistent with USP8 being a key regulatory component in the degradation of KCa3.1.

USP8 knockdown prevents lysosomal degradation and deubiquitylation of KCa3.1

To further confirm a role of USP8 in the deubiquitylation and lysosomal degradation of KCa3.1, we depleted endogeneous USP8 using siRNA. As above, the channel was streptavidin-labeled at the cell surface and degradation examined in cells transfected with either control siRNA or an siRNA directed against USP8. As shown in Fig. 6A, siRNA knockdown of USP8 expression was efficient and specific and resulted in channel degradation being significantly inhibited, with 85 ± 8% (n=3; P<0.05) of the initial membrane KCa3.1 still present 5 h later, compared to 61 ± 7% (n=3) of the initial membrane KCa3.1 still present with control siRNA, in agreement with our previously reported data (4).

Figure 6.

Knockdown of USP8 prevents KCa3.1 deubiquitylation and channel degradation. A) Cells stably expressing BLAP-KCa3.1 were transfected with siRNA control or USP8-specific siRNA. Degradation rate of the channel was significantly inhibited in USP8-depleted cells (bar graph, n=3). *P < 0.05. B) Cells were treated and labeled as in A. Ubiquitylation of KCa3.1 was assessed by GST-TUBE pulldown assay. A representative blot is shown, indicating that knockdown of the endogenous USP8 impairs deubiquitylation of internalized KCa3.1.

We next examined the effect of USP8 knockdown on KCa3.1 deubiquitylation. As shown in Fig. 6B (left panel), in cells treated with control siRNA, KCa3.1 was strongly ubiquitylated at 90 min, and ubiquitylation was markedly diminished after 5 h, indicating deubiquitylation had occurred. In cells treated with USP8 siRNA (Fig. 6B, right panel), deubiquitylation of KCa3.1 was significantly reduced. Similar results were observed in 3 experiments. Thus, USP8 is necessary for efficient deubiquitylation of KCa3.1. Taken together, these data indicate that depletion of endogenous USP8 resulted in impaired deubiquitylation of KCa3.1 and prevented lysosomal degradation of the channel.

DISCUSSION

We previously demonstrated that KCa3.1 is rapidly internalized from the plasma membrane and targeted for lysosomal degradation via an ESCRT- and Rab7-dependent pathway (4). In this study, by employing a combination of novel techniques, we determined the role of ubiquitylation in KCa3.1 endocytosis and sorting into the MVB pathway. When attempting to isolate and characterize ubiquitylated proteins from cell lysates, one of the main technical obstacles to overcome is the possible protein degradation by proteases and DUB-induced deubiquitylation of the substrates. Moreover, the classic approach of evaluating the ubiquitylation of a membrane protein is a complex and time-consuming process, which requires a random cell surface biotinylation, followed by streptavidin pulldown and subsequent immunoprecipitation of the protein of interest, and finally an IB for ubiquitin. To bypass these obstacles, in this study, we have employed a fast and more specific approach based on the newly developed BLAP-tagged KCa3.1 in combination with GST-TUBEs. BLAP-KCa3.1 allows a fast, specific biotinylation and streptavidin labeling of the channel at the cell surface (4, 14) and TUBEs enable the isolation of the ubiquitylated proteins from cell lysates while inhibiting their deubiquitylation (13). The ubiquitylated protein of interest is captured following a simple GST pulldown, which is faster and more specific than a classic IP protocol as it avoids the need for immunoprecipitating antibodies. In addition, this method is more sensitive for KCa3.1; requiring only 500 μg of total protein to detect the ubiquitylation of plasma membrane-localized channel, as compared to 2–3 mg of total protein, required by the classic IP protocol. Using this approach, we have shown that KCa3.1 is heavily ubiquitylated following endocytosis (Fig. 1B). Given the high affinity of TUBEs for polyubiquitin moieties over the single ubiquitin binding-associated domain (13), we conclude that subsequent to endocytosis the channel is tagged with polyubiquitin chains. This conclusion is also supported by the large smear of ubiquitylated KCa3.1 following endocytosis, as shown in Fig. 1C, E. It is not clear why we observe discrete bands when we pull down TUBEs or IP ubiquitin and blot for the channel (Fig. 1B, D), whereas we observe a high molecular mass smear of ubiquitylated KCa3.1 when we immunoprecipitate the channel and probe for ubiquitin (Fig. 1C, E); however, similar phenomena have been reported for other proteins (30, 31). A possible explanation for the observed smear might be the presence of other ubiquitylated proteins, which are pulled down or immunoprecipitated together with KCa3.1. The strong polyubiquitylation of KCa3.1 following internalization may serve as the sorting signal engaging the channel into the MVB-degradative pathway, mediated by ESCRT proteins. In this respect, it has been shown for other membrane proteins and receptors that polyubiquitylation at the level of the endosomal membrane functions as the essential sorting signal for protein degradation in the lysosome (32).

Further, by inhibiting channel endocytosis using hypertonic sucrose, we determined that KCa3.1 ubiquitylation may be initiated at the plasma membrane, although at very low levels (Fig. 2). The role of ubiquitylation in channel endocytosis was confirmed by treating the cells with UBEI-41, which dramatically reduced channel ubiquitylation and internalization (Fig. 2). These results suggest that endocytosis of KCa3.1 requires an intact ubiquitylation machinery. The low amount of ubiquitylated channel detected at the plasma membrane at time 0 min, which is increased under sucrose treatment, suggests that ubiquitylated KCa3.1 might itself be the internalization signal. However, we cannot rule out the possibility that the signal observed under these conditions is determined by a very low amount of internalized channel, present in a pool beneath the plasma membrane, at levels undetectable by IF. Moreover, ubiquitylation of other intermediate proteins may be required to activate the endocytotic machinery responsible for KCa3.1 internalization, similar to what has been demonstrated for other membrane proteins (33).

As ubiquitylation is a highly dynamic process, a clear understanding of ubiquitin-dependent targeting of plasma membrane KCa3.1 for lysosomal degradation requires knowledge of the role of DUBs, which act downstream of ubiquitylation and have the potential to determine the ultimate fate of internalized proteins (34). For example, deubiquitylation at an early stage of the endosomal sorting process can promote the recycling of the cargo, as opposed to increased ubiquitylation, which can act as a trigger for lysosomal degradation (27, 34). The first insight into the role of DUBs on lysosomal degradation of KCa3.1 came with the observation that treatment with a general DUBs inhibitor (PR-619) delays channel degradation rate, while maintaining the channel in a highly ubiquitylated state (Fig. 3). Thus, although KCa3.1 does require ubiquitylation at the earlier stages of the pathway for correct lysosomal sorting (Fig. 1), the channel ultimately needs to be deubiquitylated for lysosomal degradation to occur (Fig. 3).

There are more than 90 DUBs in the human genome (35, 36). Although over 90% of cellular proteins will be ubiquitylated and deubiquitylated at some point in their life cycle, very few DUB-substrate pairs have been confidently established. Approaches to do this have included using siRNA screens (26, 34), as well as tagged and purified DUBs (36, 37). Identifying a particular DUB associated with the deubiquitylation of a protein of interest is technically challenging, due to the transient nature of the enzyme-substrate interaction. Another difficulty to overcome is the fact that the ubiquitylated form of a given protein generally represents the minority fraction of the total population of that protein. This type of question is best suited to what is often termed interaction proteomics and is most commonly addressed via protein microarray, mass spectrometry, or 2-hybrid screening. Although all three have established records of success in identifying interaction partners of particular proteins, it is the microarray that is optimally suited to address the protein-protein interaction of an enzyme and substrate. Recently, Loch et al. (19) developed a DUB protein microarray that contains 35 DUBs (and ubiquitin-like isopeptidases) in addition to several deSUMOylases (SENP1, SENP2, SENP6, ULP1), a deNeddylase (DEN1), and a delISGylase (USP18). Herein, we demonstrate, for the first time, that a DUB protein array can be utilized to identify the interaction between a protein from a cell lysate and a DUB. The approach employed is based on the fact that we can specifically biotinylate and label BLAP-KCa3.1 at the cell surface with fluorophore-conjugated streptavidin, and then collect the ubiquitylated channel at different time points subsequent to endocytosis using TUBEs, followed by hybridization to the microarray. Quantification of the DUB array data (Fig. 4B) indicated that KCa3.1 associated with USP2 and USP8, and to a lesser extent with AMSH, after being internalized for 90 min at 37°C. Moreover, we do not detect this strong association for longer internalization times (>3 h), indicative of the transient nature of this interaction.

Given that USP8 has been better characterized and previously reported as one of the DUBs regulating deubiquitylation and postendocytic trafficking of other membrane proteins, such as EGFR (34), δ-opioid receptor (38) or protease-activated receptor 2 (6), we further defined the role of this DUB in lysosomal degradation of KCa3.1. We first confirmed a close association between KCa3.1 and USP8 in the same complex by co-IP (Fig. 4C). Furthermore, we demonstrate that overexpression of WT USP8 dramatically reduced the ubiquitylation of internalized KCa3.1, in good agreement with previous reports (39), whereas overexpression of catalytically inactive USP8 markedly increased the level of KCa3.1 ubiquitylation and caused retention of ubiquitylated KCa3.1 in enlarged endosomes, with which DN USP8 is also associated (Fig. 5). This observation is supported by previous studies showing that DN USP8 causes endosomal accumulation of ubiquitylated proteins, including components of the sorting machinery and also that catalytic activity of USP8 is necessary for dissociation from endosomes (40). Next, we demonstrated that overexpression of both WT or catalytically inactive USP8 dramatically inhibited channel degradation. The effect of WT USP8 expression on KCa3.1 degradation rate could be explained by the reduced level of channel ubiquitylation, which could determine an impaired recruitment of the channel into the degradation pathway in the absence of a proper sorting signal. On the other hand, DN USP8 inhibition of KCa3.1 degradation could be attributed to a diminished deubiquitylation of KCa3.1 in endosomes and reduced channel incorporation into itralumenal vesicles of MVBs. Finally, we confirmed a role for USP8 by demonstrating that siRNA-mediated knockdown of USP8 inhibited KCa3.1 deubiquitylation and degradation (Fig. 6). Overall, our results demonstrate that USP8 is required for KCa3.1 deubiquitylation and subsequent lysosomal trafficking and degradation of the channel.

By combining a series of innovative techniques, we were able to follow plasma membrane-localized KCa3.1 internalization and study its ubiquitylation in a faster and more specific manner than previously possible. To our knowledge, this is the first study demonstrating a dynamic ubiquitylation/deubiquitylation process in regulating the postendocytic trafficking and targeting for degradation of a K⁺ channel. Future studies will be carried out on USP2 to confirm a role for this DUB in the regulation of KCa3.1. Significantly, using the DUB array, we were able to identify specific interacting deubiquitylating partners for KCa3.1, with temporal and spatial resolution. That is, we only detected KCa3.1-DUB interactions in endosomes after 90 min, and this association was lost at later time points (Fig. 4B). We anticipate that DUB-protein interactions in additional cellular compartments will be amenable to this approach using an array of epitope tags. For example, proteins can be fluorescently tagged and staged in the Golgi (41), such that DUBs associated with these proteins in the Golgi and following release can be evaluated. Also, mutations that result in ER retention and/or proteasomal targeting and degradation could potentially be evaluated using HA, myc, or GFP tags with minor modifications to the protocol outlined herein. Given the critical role that DUBs play in a host of physiological processes, the ability to rapidly screen protein-DUB interactions from cell lysates represents an important advance in the ubiquitin field.