USP8 Deubiquitinates the Leptin Receptor and Is Necessary for Leptin-Mediated Synapse Formation

Abstract

Leptin has neurotrophic actions in the hippocampus to increase synapse formation and stimulate neuronal plasticity. Leptin also enhances cognition and has antidepressive and anxiolytic-like effects, two hippocampal-dependent behaviors. In contrast, mice lacking leptin or the long form of the leptin receptor (LepRb) have lower cortical volume and decreased memory and exhibit depressive-like behaviors. A number of the signaling pathways regulated by LepRb are known, but how membrane LepRb levels are regulated in the central nervous system is not well understood. Here, we show that the lysosomal inhibitor chloroquine increases LepRb expression in hippocampal cultures, suggesting that LepRb is degraded in the lysosome. Furthermore, we show that leptin increases surface expression of its own receptor by decreasing the level of ubiquitinated LepRbs. This decrease is mediated by the deubiquitinase ubiquitin-specific protease 8 (USP8), which we show is in complex with LepRb. Acute leptin stimulation increases USP8 activity. Moreover, leptin stimulates USP8 gene expression through cAMP response element–binding protein (CREB)–dependent transcription, an effect blocked by expression of a dominant-negative CREB or with short hairpin RNA knockdown of CREB. Increased expression of USP8 causes increased surface localization of LepRb, which in turn enhances leptin-mediated activation of the MAPK kinase/extracellular signal–regulated kinase pathway and CREB activation. Lastly, increased USP8 expression increases glutamatergic synapse formation in hippocampal cultures, an effect dependent on expression of LepRbs. Leptin-stimulated synapse formation also requires USP8. In conclusion, we show that USP8 deubiquitinates LepRb, thus inhibiting lysosomal degradation and enhancing surface localization of LepRb, which are essential for leptin-stimulated synaptogenesis in the hippocampus.

Leptin is a 16-kDa cytokine that is critical for normal energy homeostasis and feeding behavior (1–3). In adults, leptin is predominately synthesized and released from white adipose tissue into the blood stream, where it acts in multiple brain regions to promote negative energy balance (4, 5). The leptin receptor (LepR) is expressed in many brain regions, including the cornu ammonis (CA)1/CA3 and dentate gyrus of the hippocampus (6–8). Leptin is actively transported into the central nervous system (CNS) (9), where it binds to the long form of the LepR (LepRb) to activate multiple signaling cascades downstream of Janus kinase 2 that are important for its actions, including signal transducer and activator of transcription 3/5, MAPK kinase/extracellular signal-regulated kinase (Erk), phosphatidylinositol 3-kinase, and Ca²⁺/calmodulin-dependent protein kinase kinase/Ca²⁺/calmodulin-dependent protein kinase type I signaling pathways (10, 11).

In addition to the regulation of energy homeostasis, leptin has neurotrophic actions, both during development and in adults (10, 12–15). Mice that do not produce leptin [obese (or ob/ob) mice] have lower brain weight and lower cortical volume, a phenotype that is rescued by leptin injections during early postnatal development (12). Leptin is also essential for the formation of appropriate neuronal connections during development (13, 16). Interestingly, both the postnatal surge in leptin levels in rodents (17) and the higher leptin levels at the end of the third trimester in humans correspond to a time of rapid hippocampal synaptogenesis. Furthermore, mice lacking functional LepRbs have a reduced number of hippocampal dendritic spines, which are the main site of glutamatergic synapses (14). This reduction in hippocampal connections is associated with a change in hippocampal-dependent behaviors, as both obese (or ob/ob) mice (lacking leptin), as well as diabetic (db/db) mice (that lack functional LepRbs) exhibit increased depressive-like behaviors and anhedonia (18–20).

Leptin also has important effects on hippocampal function in adults. It has been shown to alter hippocampal synaptic function through the trafficking of glutamate receptors and enhanced long-term plasticity (21–24). Moreover, intrahippocampal injections of leptin also enhance cognition and decrease depressive and anxiety-like behaviors (25–29), whereas targeted deletion of LepRb in the adult hippocampus induces depressive-like behaviors (30).

Interestingly, mice that are heterozygous for LepRb deletion show a partial phenotype, at least for body weight (31), suggesting that the alteration of the level of functional LepRbs can have a profound influence on leptin’s actions. The alteration of the membrane expression or stability of LepRbs would therefore also be expected to impact leptin’s effects, but little is known about how protein levels and the subcellular location of LepRb are controlled, especially in neurons. LepRb expression is regulated by the ubiquitin (Ub) signaling system in immortalized cell lines, where it is constitutively endocytosed in a clathrin-mediated manner (32) and degraded in the lysosomal pathway (33). The deubiquitinase (DUB) Ub-specific protease 8 (USP8) also alters LepRb trafficking and degradation in human embryonic kidney (HEK)293T cells (34). USP8 is expressed in the CA1, CA2/CA3, and dentate gyrus of the hippocampus (35, 36), suggesting that it may play a role in regulating LepR protein levels and localization. Whereas it is expressed in the CA1 and dentate gyrus, both during the neonatal period and in adulthood, it is only expressed in the CA2/CA3 region of the hippocampus during early neonatal development (37), suggesting that it may be more involved in early postnatal development (including during the leptin surge) in these regions.

Like leptin, ubiquitinases and DUBs also play critical roles in regulating synapse formation and function (38–41). USP8 has been shown to control synaptic strength, in part, through altering α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPA) receptor expression in mammalian synapses (36), while also positively regulating dendritic spine formation (42). Given the overlapping expression of LepRbs and USP8 in the hippocampus and both their critical roles in the development and function of synapses, it raises the question of whether they interact.

Here, we show that leptin increases the expression of USP8 in the hippocampus and activates it to increase the surface expression of its own receptor (LepRb). Furthermore, we found that USP8 is in a complex with LepRb and upon activation, deubiquitinates the receptor. USP8 then increases the surface localization of LepRb, as well as leptin activation of Erk1/2 and cAMP response element–binding protein (CREB). Moreover, we found that leptin-stimulated synapse formation is dependent on USP8 expression, as targeted knockdown of USP8 blocks leptin’s effects, whereas USP8 overexpression alone is sufficient to increase synapse formation through a mechanism that requires LepRbs. This study provides evidence of how leptin activates USP8, which leads to deubiquitination of LepRbs, their increased membrane expression and leptin-induced signaling, as well as synapse formation in the hippocampus.

Materials and Methods

Drugs and DNA constructs

Full-length rat recombinant leptin (50 nM; Peprotech), MG-132 (50 µM; Selleckchem), chloroquine (50 µM; Sigma Aldrich), cycloheximide (20 µM; Sigma Aldrich), and sodium orthovanadate (1 mM; Sigma Aldrich) were used, as described in the text and figure legends. The short hairpin RNA (shRNA) targeting the USP8 sequence 5′-GGACAGGACAGTATAGATACA-3′ was cloned into the pLKO.3G vector between the EcoRI and PacI restriction sites. The constructs expressing shLepRb, shCREB, and a dominant-negative CREB (ACREB) were used as previously described (14, 43). Ub constructs were subcloned from pRK5- hemagglutinin (HA)-Ub or pRK5-HA–ubiquitin lysine 48 arginine mutant (Ub^K48R) (a kind gift from Dr. Michael Varnum, Washington State University) into the pCAGGS vector using Gateway Cloning (Thermo Fisher Scientific). Constructs expressing tagged proteins were constructed by amplification of LepRb from mouse cDNA and USP8 and β-actin from rat cDNA and cloned into pCAGGS destination vectors containing the designated tag using Gateway Cloning. Leptin receptor b lacking the intracellular domain (LepR^ΔC) was constructed by amplification of the LepRb extracellular and transmembrane-expressing domain (residues 1 to 860) from the LepRb Gateway entry vector and cloned into pCAGGS destination vectors using Gateway Cloning. The catalytically inactive USP8 [ubiquitin-specific protease 8 cysteine 749 alanine mutant (USP8^C749A)] construct was mutated with primers designed by NEBaseChanger. The USP8 construct resistant to shUSP8 (V5-USP8*) was constructed silently by the mutation of the shRNA targeting sequence with primers designed by NEBaseChanger. BioID destination vectors were constructed by the subcloning of myc-BioID (Addgene) into pCAGGS destination vectors using Gibson cloning (New England Biolabs). Primers are reported in Bland et al. (44).

Cell culture

All animal work described in this manuscript followed procedures approved and in compliance with Washington State University Institutional Animal Care and Use Committee–approved protocols. Sexually-mixed hippocampal neuronal cultures were prepared from equal numbers of male and female pups, as previously described (10). Hippocampal neurons used for immunostaining and fluorescent intensity imaging were transfected with various constructs on day 6 in vitro (DIV6) and then on DIV7 to -8, treated with various reagents in media, fixed [4% paraformaldehyde in PHEMS buffer (60 mM 1,4-piperazinediethanesulfonic acid, 25 mM HEPES, 1 mM MgCl₂, 5 mM EGTA), 87.6 mM sucrose, pH 7.4] for 20 minutes at room temperature, and then immunostained or mounted using Elvanol. Hippocampal neurons used for dendritic spine analysis, antivesicular glutamate transporter 1 (vGlut1) immunostaining, and electrophysiological analysis were transfected with various constructs on DIV6 and treated with leptin (50 nM) on DIV8, and on DIV11 to -12, cells were fixed, immunostained, and mounted, as previously stated, or used for electrophysiological recordings. All control conditions received the same amount of media at the time of reagent stimulation.

HEK293T cells (5 × 10⁴ cells/cm² for six-well plates used for biochemistry experiments) were maintained in DMEM/high glucose (HyClone), supplemented with 10% fetal bovine serum (Atlanta Biologicals) and 1% penicillin/streptomycin (Gibco). HEK293T cells were transfected with various constructs, 24 hours after plating, treated 24 to 48 hours after transfection with leptin (50 nM) in Opti-Mem (Gibco), and collected at designated time points. All control conditions received the same amount of media at the time of reagent stimulation.

Transfection

Primary hippocampal cultures were transfected with Lipofectamine 2000 (Life Technologies). Native media was collected before transfection and replaced with warm growth media. Lipofectamine 2000 and experimental DNA plasmids (0.5 µg per well for 24-well plates, 2 µg per well for six-well plates, and 12.5 µg/dish for 10 cm dishes) were added to cells and incubated for 30 minutes. The media was then aspirated and replaced with native media. This protocol produces a transfection efficiency of only 3% to 5% of total neurons transfected.

HEK293T cells were transfected with Lipofectamine 2000 and experimental DNA plasmids (2 µg per well for six-well plates). The mixture was added to the native media and allowed to incubate until cells were collected. This protocol produces a much higher transfection efficiency of ∼80% of total cells transfected.

Microscopy

In all of the fluorescent intensity experiments, a soluble fluorescent protein, such as tdTomato or Clover, was used to highlight the transfected neuron. Wide-field images of transfected neurons were obtained using Slidebook 5.5 Digital Microscopy Software and an Olympus IX81 inverted microscope equipped with a Hamamatsu ORCA-ER charge-coupled device camera and a 100× oil immersion lens (numerical aperture: 1.4). The soluble fill was then used to mask the transfected neuron. The masked image was then used to identify the region of interest to be analyzed (the entire neuron) for LepRb expression. Total fluorescent integrated intensity for the entire neuron was then measured.

Immunocytochemistry

Transfected neurons were treated and fixed, as described above. After fixation, cells were washed in PBS and permeabilized with 0.1% Triton X-100 detergent (Bio-Rad Laboratories); blocked with 8% BSA for 1 hour; incubated for 24 hours at 4°C with primary antibodies against USP8 (1:250; Sigma Aldrich) (45), phosphorylated (p)Erk (1:250; Cell Signaling Technology) (46), pCREB (1:250; Millipore) (47), or vGlut1 (1:250; Neuromab) (48) in IHC-Tek (IHC World); incubated for 1 hour at room temperature with the appropriate Alexa Fluor secondary IgG antibody (Life Technologies); and mounted with Elvanol. Antibodies were all validated for specificity by Western blot of HEK-expressed target protein or with no primary control in immunocytochemical experiments. The integrated staining density of pErk was measured in the whole soma, and pCREB was measured in the nucleus of transfected neurons from a minimum of 30 neurons from three separate cultures per condition using ImageJ 1.48. Images were taken as described in the Microscopy section. Background was calculated using secondary antibody incubation only and subtracted from images.

Spine quantification

Hippocampal cultures were transfected with designated DNA constructs and Clover-β-actin to allow for visualization of dendritic spine density and morphology. Confocal fluorescent images were obtained using Metamorph software and a Leica DMI6000 SD confocal microscope equipped with a Yokogawa CSU-X1 spinning disk, Hamamatsu-R2 charge-coupled device camera, and a 60× oil immersion lens (numerical aperture: 1.4). Dendritic spine density and classification were measured, as previously described (10). In summary, two to three dendritic segments were counted from a minimum of 15 neurons. Confocal images for vGlut1 and dendritic spine juxtaposition were obtained, as described earlier. The percent juxtaposition was manually measured from 50 to 75 spines on 10 different neurons from three separate cultures per condition using ImageJ 1.48.

Live immunostaining of surface Flag-LepRb was performed on neurons transfected and treated, as previously described. Neurons were incubated in growth media, supplemented with an anti-Flag (1:250; Sigma Aldrich) (49) antibody for 10 minutes, fixed, and incubated with the appropriate Alexa Fluor secondary IgG antibody for 1 hour at room temperature. They were then mounted, imaged, and analyzed, as described in the Microscopy section. Background was calculated using secondary antibody incubation only and subtracted from images.

Western blotting

Protein samples were collected by the lysing of equal numbers of cells or equal masses of tissue with radioimmunoprecipitation assay buffer (Santa Cruz Biotechnology); supplemented with phosphatase inhibitor cocktail 2 and 3 (Sigma Aldrich), protease inhibitor cocktail, phenylmethylsulfonyl fluoride (2 mM), sodium orthovanadate (1 mM; Santa Cruz Biotechnology), and PR-619 (25 µM; Tocris); and centrifuged at 16,000 g. Samples were prepared with NuPage LDS Sample Buffer (Life Technologies) and 50 mM dithiothreitol (DTT; Thermo Fisher Scientific) and heated at 75°C for 10 minutes. Equal concentrations of protein were then loaded into Bolt 4% to 12% Bis-Tris gels (Life Technologies). Proteins were transferred to a polyvinylidene fluoride membrane (Life Technologies) overnight; blocked with 5% BSA for phosphorylated proteins or 5% milk for nonphosphorylated proteins; incubated with primary antibodies against c-Myc (1:1000; Sigma Aldrich) (50), HA (1:1000; Sigma Aldrich) (51), Erk2 (1:1000; Santa Cruz Biotechnology) (52), USP8 (1:1000; Sigma Aldrich) (45), pTyr (1:500; MilliporeSigma) (53), or V5 (1:1000; Cell Signaling Technology) (54) for 2 hours at room temperature; and then incubated with the appropriate Alexa Fluor-647 secondary IgG F(ab′)₂ fragment antibody (Cell Signaling Technology) (55) for 1 hour at room temperature. Depending on the experiment, loading controls included Erk2 or the protein being analyzed [in the case of immunoprecipitation (IP) and phosphoblots]. Blots were imaged using a Chemidoc MP imaging system (Bio-Rad Laboratories) and analyzed using the ImageJ 1.48 gel analyzer tool.

IP

Protein samples were collected as described in the “Western blotting” protocol (above) from cells transfected with the designated constructs. Equal amounts of sample were incubated with myc-tag or V5-tag mouse monoclonal antibody conjugated to magnetic beads (Cell Signaling Technology) and rotated overnight at 4°C. Samples were then washed three times with PBS, eluted in NuPage LDS Sample Buffer and 50 mM DTT, heated at 75°C for 10 minutes, and then loaded into Bolt 4% to 12% Bis-Tris gels. Samples were probed in Western blot analysis, as previously described.

Active DUB labeling assay

The activity of USP8 was measured, as previously described (36). In summary, USP8 was immunoprecipitated, as described in the IP section protocol, from neurons transfected with myc-USP8. Before elution from magnetic beads, samples were incubated with the active DUB labeling HA-Ub-vinyl methylester (VME; 1 µM; Enzo Life Sciences) for 1 hour at 37°C. Samples were then eluted in NuPage LDS Sample Buffer and 50 mM DTT and probed in Western blot analysis.

BioID assay

HEK293T cells were transfected with a V5-tagged target protein and a BioID-tagged labeling protein. Biotin (50 µM; Amresco) was added to the media at the time of transfection to allow the BioID-tagged labeling protein to attach biotin to any proteins in close proximity. Protein samples were collected after 24 hours, as described in the “Western blotting” protocol (above), and equal volumes of sample were incubated with 2 µL anti-V5 antibody (Cell Signaling Technology) (54) and rotated overnight at 4°C, after which Magne Protein G beads (Promega) were added for an additional 2 hours at room temperature to immunoprecipitate the V5-tagged target protein. Samples were washed in PBS, eluted in NuPage LDS Sample Buffer and 50 mM DTT, and probed in Western blot analysis. Biotinylated proteins were stained with streptavidin, conjugated to Alexa Fluor-647 (Life Technologies) (56), and normalized to the total amount of V5-tagged protein immunoprecipitated.

Real-time quantitative RT-PCR

RNA was isolated from hippocampal cultures stimulated with leptin using the PureLink RNA Mini Kit (Invitrogen) and converted to cDNA using the SuperScript VILO cDNA Synthesis Kit (Invitrogen) with on-column DNA digest using the PureLink DNase Set (Invitrogen), according to the manufacturer’s protocols. All quantitative RT-PCRs (qRT-PCRs) were carried out using Fast SYBR Green Master Mix (Applied Biosystems), according to the manufacturer’s protocols, and run on a Viia 7 System (Applied Biosystems). All primers used in our studies are listed in Bland et al. (44). Quantification of USP8 mRNA was calculated as described in Pfaffl (57) and normalized to the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Whole-cell recordings

Whole-cell recordings were performed, as previously described in Dhar et al. (14), on DIV11 to -12 hippocampal cultures, transfected with Clover-β-actin and designated constructs.

Chromatin IP and chromatin IP-sequencing

Chromatin IP (ChIP) and ChIP-sequencing (ChIP-Seq) were performed as previously described in Lesiak et al. (58).

Statistical analysis

Electrophysiological data are expressed as means ± SE and were analyzed using a one-way ANOVA and a Tukey post hoc test. All other data are expressed as the means ± SEM. Multiple comparisons were analyzed using one-way ANOVA with post hoc Tukey analysis, and single comparisons were analyzed using a two-tailed, unpaired Student t-test. Statistical significance was set to a minimum of P < 0.05.

Results

LepRb is targeted for degradation by the lysosomal pathway in hippocampal neurons

Leptin requires the surface expression of LepRb for activating downstream signaling cascades important in glutamatergic synaptogenesis (10, 14). Surface expression of LepRb has been demonstrated to be regulated by endocytosis via both ligand-dependent or constitutive mechanisms. Moreover, studies in heterologous cell lines have demonstrated that LepRb can be internalized and degraded by either the lysosomal or the proteasome pathways (32–34). These studies suggest that there are cell type-specific mechanisms for the regulation of LepRb surface expression. The mechanism by which surface expression of LepRb is regulated in neurons is unknown. To begin to identify the mechanism(s) by which LepRb expression is regulated in hippocampal neurons, we tested if LepRb is targeted for lysosome or proteasome-mediated degradation. We transfected cultured hippocampal neurons with a carboxyl-terminal, fluorescently tagged LepRb (LepRb-Clover) and tdTomato as a soluble cellular marker; then treated the cultures with either the proteasome inhibitor MG-132 (50 µM) or the lysosomal inhibitor chloroquine (50 µM); and measured the fluorescent intensity of LepRb-Clover (Fig. 1A). Neuronal cells were masked using the tdTomato channel, and the integrated intensity of the LepRb-Clover channel was measured. Inhibition of the lysosomal degradation pathway increased LepRb-Clover fluorescent intensity within 2 hours, which is further enhanced at 4 hours (2 hours: 140% ± 10%, P < 0.05; 4 hours: 148% ± 14%, P < 0.05), whereas inhibition of the proteasome does not affect LepRb-Clover fluorescent intensity (Fig. 1B).

Figure 1.

Leptin stimulation increases LepRb expression. (A) Representative fluorescent images of hippocampal cultures expressing tdTomato and LepRb-Clover. Cultures were stimulated with MG-132 (50 µM), chloroquine (50 µM), leptin (50 nM), cycloheximide (20 µM), or a combination of cycloheximide and leptin the next day for 2 h or 4 h and then fixed and imaged. Original scale bars, 25 µm. (B) Quantification of LepRb-Clover intensity normalized to control condition (dotted line; n = 30 to 49 neurons). The experiment was repeated in three independent hippocampal culture preparations. (C) qRT-PCR analysis of LEPRB mRNA levels in hippocampal neurons treated with leptin (50 nM). LEPRB transcript levels were normalized to GAPDH. (D) Representative Western blot (WB) of HEK293T cells expressing LepRb-myc and HA-Ub. Cells were treated with leptin (50 nM, 4 h) and lysed with lysis buffer containing a deubiquitinating enzyme inhibitor (PR-619; 25 μM), and LepRb-myc was immunoprecipitated. (E and F) Quantification of total (E) LepRb-myc intensity in the input lane normalized to Erk2 intensity or (F) ubiquitinated LepRb-myc intensity in the IP lane normalized to total LepRb-myc intensity in the same lane (n = 4). All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01 compared with control.

Leptin decreases the degradation of LepRb

Next, we show that acute leptin (50 nM) stimulation of hippocampal cultures increases LepRb-Clover intensity within 2 hours, which is further enhanced at 4 hours (2 hours: 138% ± 12%, P < 0.05; 4 hours: 155% ± 13%, P < 0.01; Fig. 1B). We also observed that the expression level of exogenously expressed LepRb-myc in HEK293T cells is enhanced (251% ± 44%, P = 0.0259) within 4 hours of leptin stimulation (Fig. 1D and 1E).

Increased expression of LepRb could be the result of increased transcription, translation, or protein stability. To test if leptin stimulation increases transcription of the LepRb gene, we monitored mRNA levels following leptin stimulation. We saw no substantial change in mRNA encoding LepRb up to 8 hours following leptin stimulation using qRT-PCR analysis (Fig. 1C).

To distinguish the role of increased protein synthesis vs degradation of LepRb with leptin stimulation, we treated hippocampal cultures expressing LepRb-Clover with leptin alone, the protein synthesis inhibitor cycloheximide (20 µM), or a combination of leptin and cycloheximide (Fig. 1A). As expected, cycloheximide treatment by itself decreases LepRb-Clover fluorescent intensity (2 hours: 90% ± 9%; 4 hours: 54% ± 4%, P < 0.05), which was returned to near-control levels with leptin treatment (2 hours: 124% ± 13%, P < 0.01; 4 hours: 166% ± 8%, P < 0.01 compared with cycloheximide alone; Fig. 1B), suggesting that leptin’s effect to increase surface expression of LepRbs is independent of synthesis.

Leptin reduces its own degradation through ubiquitination in HEK293T cells

The attachment of Ub to proteins can be used to mark proteins for both lysosome and proteasome degradation. To determine whether LepRs can be regulated in this manner, we first looked at the effect of leptin on the levels of Ub-LepRb in the HEK293T cell expressing LepRb-myc and HA-Ub and found that leptin decreases the levels of Ub-LepRb-myc (56% ± 9.4%, P = 0.019; Fig. 1D and 1F). These data suggest that leptin activation of LepRb reduces LepRb degradation through a deubiquitination-mediated mechanism in cell lines. We therefore wanted to determine whether leptins also alter deubiquitination of LepRb in hippocampal neurons. To begin to address this, we first examined the effect of leptin on deubiquitinating enzymes in hippocampal neurons.

Leptin enhances USP8 gene expression

We have previously shown that the CREB is activated by leptin (14). Interestingly, CREB Chip-Seq experiments identified DUB USP8 as one gene with CREB bound in its promotor region (Fig. 2A). qRT-PCR experiments show USP8 mRNA begins to increase in hippocampal cultures within 2 hours of leptin stimulation and remains elevated for >24 hours (2 hours: 129% ± 13%; 4 hours: 131% ± 5.4%, P < 0.01; 8 hours: 148% ± 16.3%, P < 0.05; 24 hours: 133% ± 10%, P < 0.05; Fig. 2B). Immunostaining of endogenous USP8 in hippocampal neurons transfected with soluble Clover to demark neuronal bodies and projections also shows elevated protein levels at 4 hours of leptin stimulation in hippocampal cultures (155% ± 13%, P < 0.01; Fig. 2C and 2D). This was performed by the masking of the neuronal cells using the Clover channel and measurement of the integrated intensity of the immunostained USP8 channel. Expression of ACREB or shCREB blocks the leptin-induced increase in USP8 expression (Fig. 2C and 2D), supporting that the hypothesis leptin regulates USP8 via a CREB-dependent pathway. Interestingly, the expression level of USP8 is significantly decreased (41% ± 3.8%, P = 0.027) in leptin signaling-deficient mice that lack the LepRb (db/db mice; Fig. 2E and 2F), suggesting that under normal conditions, leptin activation of LepRbs controls the expression of USP8 in the hippocampus.

Figure 2.

Leptin induces USP8 gene transcription. (A) The University of California, Santa Cruz, genome browser tracks show hippocampal CREB ChIP-Seq, embryonic cortical neuron CREB ChIP-Seq (Neocortical CREB) data relative to Reference Sequence genes, and CpG islands (green). (B) qRT-PCR analysis of USP8 mRNA levels in hippocampal neurons treated with leptin (50 nM). USP8 transcript levels were normalized to GAPDH. (C) Representative fluorescent images of hippocampal neurons expressing Clover, stimulated with leptin (50 nM), and immunostained for USP8. Original scale bar, 25 µm. (D) Quantification of USP8 integrated signal density normalized to control condition (n = 30 neurons). (E) Representative Western blot of P10 wild-type (wt) and db/db mice hippocampal protein extracts. (F) Quantification of USP8 intensity normalized to Erk2 intensity (wt: n = 4; db/db: n = 5). All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01 compared with control.

Leptin increases USP8 activity in hippocampal neurons

Phosphorylation and dephosphorylation of USP8 at critical tyrosine (Tyr) residues can either increase or decreases DUB activity of USP8, depending on which residues are phosphorylated (36, 59). We therefore first tested whether leptin can alter pUSP8 Tyr residues by IP of USP8 from HEK293T cells expressing a myc-USP8 (and LepRb) and measurement of the level of myc-USP8 pTyr (Fig. 3A). Leptin decreases (67% ± 11%, P = 0.05) the level of pTyr myc-USP8 within 1 hour (Fig. 3B). USP8 produces two bands on Western blots with the dephosphorylated form of USP8 migrating faster in an SDS-PAGE gel (36).

Figure 3.

Leptin stimulation decreases USP8 pTyr. (A) Representative Western blot of myc-USP8 immunoprecipitated from HEK293T cells expressing myc-USP8 and stimulated with leptin (50 nM). (B) Quantification of myc-USP8 pTyr normalized to total myc-USP8 (n = 3). (C) Representative Western blot of hippocampal neurons treated with leptin (50 nM). (D) Quantification of the ratio of the active non-pUSP8 (bottom band) to the nonactive pUSP8 (top band), normalized to control condition (n = 3). (E) Representative Western blot from hippocampal neurons treated with leptin (50 nM) or leptin and Na₃VO₄ (1 mM) for 1 h. Designated lysates were treated with λ protein phosphatase (λPP) for 1 h at 30°C. (F) Quantification of the ratio of the active non-pUSP8 (bottom band) to the nonactive pUSP8 (top band), normalized to control condition (n = 3). (G) Representative Western blot of myc-USP8 immunoprecipitated from hippocampal neurons expressing myc-USP8 and treated with leptin (50 nM, 1 h). Immunoprecipitated myc-USP8 was incubated with HA-Ub-VME (1 µM) for 1 h at 37°C. (H) Quantification of HA-Ub-VME-USP8 intensity normalized to total myc-USP8 intensity in the same lane (n = 4). All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01 compared with control condition. NS, not significant.

Next, to determine if leptin also decreases the endogenous pUSP8 in hippocampal cultures, we measured the ratio between the faster migrating USP8 band (dephosphorylated USP8) and the slower migrating band (pUSP8; Fig. 3C). Leptin increases (1.29 ± 0.01, P < 0.05) the USP8/pUSP8 ratio within 60 minutes (Fig. 3D). To determine further if this shift in molecular weight is a result of dephosphorylation, we treated cellular lysates with λ protein phosphatase (λPP). Treatment with λPP greatly increased [λPP: 2.58 ± 0.3, P < 0.01; λPP + leptin (1 hour): 3.16 ± 0.4, P < 0.01] the USP8/pUSP8 ratio in both control and leptin-stimulated cultures, whereas treatment with the pTyr inhibitor Na₃VO₄ (60) inhibited leptin dephosphorylation of USP8 (Fig. 3E and 3F).

As others have shown that dephosphorylation of USP8 can enhance its activity (36), we next wanted to determine whether leptin increases the activity of USP8 in hippocampal neurons. To examine this, we immunoprecipitated USP8 from hippocampal cultures expressing myc-USP8 and treated with leptin for 1 hour (Fig. 3G). The immunoprecipitated myc-USP8 was then labeled with the DUB activity probe HA-Ub-VME which will only bind to active USP8 (61). Leptin increases (161% ± 10%, P < 0.05) the binding of HA-Ub-VME to myc-USP8 (Fig. 3H), suggesting increased USP8 activity with leptin stimulation. Taken together, these data show that leptin both decreases pTyr of USP8 and enhances its activity in hippocampal neurons.

USP8 is in a complex with LepRb

One challenge in finding the targets of DUBs is determining if they deubiquitinate the target protein themselves or if the deubiquitination occurs through an intermediate DUB. To test if USP8 is associated with, or in close proximity to, LepRb, we used a BioID method (62) combined with co-IP assays. As a result of poor transfection efficiency and very low expression of LepRb-expressing constructs in hippocampal neurons, we performed these biochemical analyses in HEK293T cells, which show a more robust expression of LepRb using these same constructs. First, to determine if USP8 is in close proximity to LepRb, we used BioID (Fig. 4A). This method uses a construct that fuses the promiscuous biotin ligase myc-BirA* (which will be further listed as BioID) to a protein of interest, which can then add biotin to any proteins that are in close proximity to the BioID-tagged protein (≤20 nm labeling radius) (63). We first expressed a LepRb-BioID fusion protein or a truncated form of LepRb (LepR^ΔC-BioID), which only includes the extracellular and transmembrane domain of LepRb (residues 1 to 860), to normalize for nonspecific biotinylation. In the presence of biotin, LepRb-BioID will covalently attach biotin to proteins that are either in a complex with LepRb-BioID or in close proximity. As LepR^ΔC-BioID has no intracellular domain to form complexes with other proteins, this construct was used to normalize for nonspecific biotinylation by the BioID tag. To determine if USP8 is in a complex with LepRb, we coexpressed LepRb-BioID and V5-USP8 in HEK293T cells and incubated them for 24 hours in media supplemented with biotin. To measure the ability of LepRb-BioID to biotinylation V5-USP8, we immunoprecipitated V5-USP8 and measured its biotinylation levels [Fig. 4B; see Bland et al. (44)]. LepRb-BioID robustly biotinylates V5-USP8 (12.1 ± 3.4-fold over LepR^ΔC-BioID, P = 0.032), suggesting that USP8 is localized in close proximity to LepRb but not LepR^ΔC-BioID (Fig. 4C). To determine further the proximity of these two proteins, we performed the reciprocal experiment by attaching BioID to USP8 and measuring its ability to biotinylate LepRb. This was performed by coexpression of BioID-USP8 and LepRb-V5, IP of LepRb-V5, and measurement of its level of biotinylation. As USP8 is not membrane bound, nonspecific biotinylation by BioID was normalized by the expression of a soluble form of BioID alone (BirA*). BioID-USP8 robustly biotinylates LepRb-V5 compared with soluble BirA* alone (5.7 ± 0.8-fold over BirA*, P = 0.0021; Fig. 4D and 4E). We also show that under basal conditions, V5-USP8 coimmunoprecipitates with LepRb-myc, and LepRb-myc coimmunoprecipitates with V5-USP8 in HEK293T cells (Fig. 4F and 4H); the interaction is enhanced (163% ± 15.3%, P = 0.0397, and 373% ± 36.6%, P = 0.009, respectively) following leptin stimulation (Fig. 4G and 4I). This suggests that under basal conditions, there is some USP8 bound to LepRb, and leptin stimulation promotes more binding of USP8. This further supports the hypothesis that LepRb and USP8 can occur in a complex.

Figure 4.

USP8 is in a complex with LepRb. (A) Diagram of BioID experiment. (B) Representative Western blot of V5-USP8 immunoprecipitated from HEK293T cells expressing the designated BioID constructs and V5-USP8. Biotin (50 μM) was added at time of transfection. (C) Quantification of IP biotinylated V5-USP8 normalized to total V5-USP8 in the same lane (n = 3). (D) Representative Western blot of LepRb-V5 immunoprecipitated from HEK293T cells expressing either BirA* or BioID-USP8 and LepRb-V5. Biotin (50 μM) was added at time of transfection. (E) Quantification of IP biotinylated LepRb-V5 normalized to total LepRb-V5 in the same lane (n = 3). (F–H) Representative Western blot of HEK293T cells expressing LepRb-myc and V5-USP8. Anti-IgG, (F) anti-myc, and (H) anti-V5 antibody-linked magnetic beads were used to coimmunoprecipitate LepRb-myc and V5-USP8 ± leptin stimulation (50 nM, 4 h). (G and I) Quantification of (G) coimmunoprecipitated V5-USP8 intensity normalized to LepRb-myc intensity and (I) coimmunoprecipitated LepRb-myc intensity normalized to V5-UPS8 intensity in the same lane (n = 3). All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01 compared with control.

USP8 deubiquitinates LepRb

As USP8 occurs in a complex with LepRb, and leptin stimulation increases USP8 expression and activation while decreasing LepRb ubiquitination, we tested if USP8 deubiquitinates LepRbs. To measure the level of Ub-LepRb, we expressed LepRb-myc and HA-Ub in HEK293T cells, immunoprecipitated LepRb-myc, and measured its level of ubiquitination by Western blot [Fig. 5A; see Bland et al. (44)]. Coexpression with V5-USP8 decreases (35% ± 1.2%, P < 0.01) ubiquitinated LepRb-myc levels, which enhances leptin-stimulated deubiquitination of LepRb (25% ± 2.1%, P < 0.05; Fig. 5B). We show that shUSP8 greatly decreases USP8 levels [48% ± 5%, P < 0.001; see Bland et al. (44)]. LepRb-myc shows increased levels of ubiquitination when coexpressed with shUSP8 (124% ± 6.3%, P = 0.0195; Fig. 5C and 5D). We also show that the catalytically inactive USP8 mutant (V5-USP8^C749A) (64) does not decrease basal levels of Ub-LepRb-myc and inhibits the decrease of LepRb-myc ubiquitination mediated by leptin (Fig. 5E and 5F). To demonstrate further that LepRb is ubiquitinated, we immunoprecipitated HA–ubiquitin wild type (Ub^wt) and then blotted for LepRb-Myc. Under both control and leptin-stimulated conditions, immunoprecipitated Ub coimmunoprecipitated a high molecular smear of LepRb-Myc [see Bland et al. (44)]. No significant change in the amount of LepRb-Myc coimmunoprecipitated was seen following leptin stimulation. This is most likely caused by the fact that a single molecule HA-Ub conjugated to LepRb would be sufficient to allow for IP of LepRb-Myc.

Figure 5.

USP8 deubiquitinates LepRb. (A) Representative Western blot of LepR-myc immunoprecipitated from HEK293T cells expressing LepRb-myc, HA-Ub^wt, or HA-Ub^K48R and V5-USP8 ± leptin stimulation (50 nM, 4 h). (B) Quantification of IP Ub^wt-LepRb-myc and Ub^K48R-LepRb-myc intensities normalized to total LepRb-myc in the same lane (n = 4). (C) Representative Western blot of LepR-myc immunoprecipitated from HEK293T cells expressing LepRb-myc, HA-Ub^wt, and shUSP8. (D) Quantification of IP Ub^wt-LepRb-myc intensity normalized to total LepRb-myc in the same lane (n = 3). (E) Representative Western blot of LepR-myc immunoprecipitated from HEK293T cells expressing LepRb-myc, HA-Ub^wt, and V5-USP8^C749A ± leptin stimulation (50 nM, 4 h). (F) Quantification of IP Ub^wt-LepRb-myc intensity normalized to total LepRb-myc in the same lane (n = 3). All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01 compared with control. NS, not significant; wt, wild type.

We next determined what type of ubiquitination USP8 removes from LepRb. Ub attachment can occur as a monomer or as branched poly-Ub chains. Ub that is branched from the lysine 48 residue (K48) of Ub generally targets proteins for degradation, whereas mono-Ub can signal proteins for internalization and sorting (65). To determine if USP8 can remove K48-branched Ub from LepRb, which would protect the receptor from degradation, we used a mutant form of Ub lacking the K48 branch site (Ub^K48R; Fig. 5A). When LepRb-myc is ubiquitinated with Ub^K48R, V5-USP8-mediated deubiquitination of LepRb-myc is inhibited but not fully abolished (Ub^wt: 65% ± 1.2%, P < 0.001; Ub^K48R: 39% ± 8.9%, P = 0.0299; Fig. 5B). This inhibition suggests that USP8 targets K48-branched Ub bound to LepRb. As Ub^K48R did not completely abolish deubiquitination of LepRb by USP8, this suggests that USP8 also has the ability to remove multiple forms of Ub from LepRb, other than K48-branched chains. Leptin-stimulated decrease in Ub-LepRb-myc is also inhibited with Ub^K48R expression (Fig. 5B), suggesting that leptin-mediated LepRb deubiquitination mainly involves the removal of K48-branched Ub chains.

USP8 enhances LepRb surface localization and leptin signaling in hippocampal neurons

We next determined the functional significance of USP8 deubiquitination of LepRb. Under basal conditions, there are low levels of surface-localized LepRb resulting from the majority of LepRb being localized to the Golgi complex or in a post-Golgi intracellular compartment (66). Leptin stimulation, as well as USP8 overexpression, increases total LepRb expression [Fig. 1A; see Bland et al. (44)], but to determine if they increase the surface pool of receptors, we expressed LepRb that carries an extracellular tag on its amino terminus (Flag-LepRb) in hippocampal cultures and performed live immunostaining for surface Flag-LepRb (Fig. 6A). Four-hour leptin stimulation, as well as coexpression of V5-USP8, increases surface Flag-LepRb localization, whereas shUSP8 inhibits a leptin-stimulated increase in surface LepRb localization (leptin: 223% ± 25%, P < 0.05; V5-USP8: 236% ± 40%, P < 0.01; V5-USP8 + leptin: 285% ± 41%, P < 0.01; shUSP8: 114% ± 18%; shUSP8 + leptin: 132% ± 18%; Fig. 6B). Leptin stimulation and USP8 overexpression also increased the number of Flag-LepRb surface puncta, which was blocked with shUSP8 (data not shown).

Figure 6.

USP8 increases surface expression of LepRb and enhances leptin-stimulated signaling cascade activation. (A) Representative fluorescent images of hippocampal neurons expressing Clover, Flag-LepRb, and either V5-USP8 or shUSP8 ± leptin stimulation (50 nM, 4 h). Cultures were live immunostained for surface Flag-LepRb. (B) Quantification of the Flag-LepRb integrated signal density normalized to control condition (n = 28 to 30 neurons). (C and E) Representative fluorescent images of hippocampal neurons expressing Clover and either V5-USP8 or shUSP8 ± leptin stimulation (50 nM, 15 min). Neurons were immunostained for (C) pErk or (E) pCREB. (D and F) Quantification of the (D) pErk integrated signal density or the (F) pCREB integrated signal density normalized to control condition (n = 30 neurons). All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01. Original scale bars, 25 µm.

We next tested the functional consequences of the increased surface LepRb localization. Increased surface expression of LepRb could result in either increased sensitivity or increased amplitude of the effect of leptin stimulation in the neuron. Our laboratory has previously reported that leptin activates pErk1/2 and pCREB but not phosphorylation of signal transducer and activator of transcription 3 in hippocampal cultures, an effect dependent on LepRb expression (14). To test for increased amplitude of leptin’s effects, we stimulated hippocampal cultures with leptin for 15 minutes and immunostained for pErk1/2 (pErk) and pCREB. Leptin increases pErk and pCREB levels, which is further enhanced with expression of V5-USP8 (pErk: leptin: 205% ± 24%, P < 0.01; V5-USP8: 89% ± 13%; V5-USP8 + leptin: 301% ± 28%, P < 0.05 compared with control + leptin; pCREB: leptin: 126% ± 3%, P < 0.01; V5-USP8: 92% ± 4%; V5-USP8 + leptin: 148% ± 6%, P < 0.05 compared with control + leptin), and inhibited or completely abolished shUSP8 (Fig. 6C–6F). To test whether USP8 increased sensitivity to leptin, we performed a leptin dose response ± USP8 expression and then measured Erk activation. Although UPS8 expression increased the amplitude, as described above, its expression did not shift the sensitivity of LepRb to different leptin concentrations [Bland et al. (44)]. These data suggest that USP8 expression is necessary for the surface localization of LepRbs, and USP8 expression amplifies LepRb activation of downstream signaling cascades by leptin in hippocampal neurons.

USP8 increases synapse formation in hippocampal neurons

Leptin increases the number of functional excitatory glutamatergic synapses in hippocampal neurons, an effect attenuated by LepRb shRNA knockdown (14). The alteration of surface expression of LepRb could therefore impact synaptogenesis in the hippocampus. Thus, we tested the effect of USP8 on leptin-stimulated synaptogenesis. To visualize and measure the density of dendritic spines and spine morphology, hippocampal cultures were transfected on DIV6 with a fluorescently tagged actin (Clover-β-actin), which is enriched in dendritic spines, treated with leptin on DIV8, and fixed and imaged on DIV11 (Fig. 7A). We show that USP8 overexpression in hippocampal neurons increases mature mushroom-shaped dendritic spine density (mushroom spine density/50 μm ± SEM; control: 19.3 ± 2.1; control + leptin: 27.7 ± 1.9, P < 0.05; V5-USP8: 33.3 ± 2.3, P < 0.01; V5-USP8 + leptin: 27.9 ± 1.6, P < 0.05; Fig. 7B). Interestingly, USP8 overexpression did not enhance leptin-stimulated dendritic spine formation but actually occluded it. These data might suggest some unknown homeostatic regulation of synaptic numbers following USP8 overexpression. However, enhancement of mushroom-shaped spine density by USP8 is blocked with shLepRb (mushroom spine density/50 μm ± SEM; control: 19.3 ± 2.1; control + leptin: 27.7 ± 1.9, P < 0.05; V5-USP8 + shLepR: 15.6 ± 1.2; V5-USP8 + shLepR + leptin: 15.2 ± 1.4), suggesting that LepRbs are required for the USP8 increase in dendritic spines. Moreover, we showed that leptin cannot induced dendrite spine formation following shRNA-mediated knockdown of USP8, an effect that can be rescued by expression of a shRNA-resistant USP8 mutant (V5-USP8*; mushroom spine density/50 μm ± SEM; control: 19.3 ± 2.1; control + leptin: 27.7 ± 1.9, P < 0.05; shUSP8: 15.1 ± 1.5; shUSP8 + leptin: 14.3 ± 1.4; shUSP8 + V5-USP8*: 43.3 ± 2.4, P < 0.01; shUSP8 + V5-USP8* + leptin: 37.7 ± 1.8; P < 0.01; Fig. 7B).

Figure 7.

USP8 regulates dendritic spine formation in hippocampal neurons. (A–D) DIV6 hippocampal neurons were transfected with a fluorescent Clover-β-actin and the indicated constructs. Neurons were stimulated with leptin (50 nM) on DIV8 and fixed on DIV11 to DIV12 for spine density and presynaptic and postsynaptic juxtaposition experiments. (A) Representative fluorescent images of dendrite segments. Dendritic spine density was measured by hand using ImageJ with the NeuronJ plugin. (B) Quantification of dendritic spine density (n = 15 neurons). (C) Representative fluorescent images of hippocampal cultures immunostained for the presynaptic marker vGlut1. (D) Quantification of juxtaposition of dendritic spines and vGlut1 puncta averaged from 50 to 75 spines from 10 neurons. All experiments were repeated in three independent culture preparations and presented as the means ± SEM. *P < 0.05, **P < 0.01 compared with control. Original scale bars, 5 µm.

To determine if USP8 induces an increase in functional synapses, we measured the percent of dendritic spines that are juxtaposed to the presynaptic marker vGlut1 (Fig. 7C), and we also measured the frequency, amplitude, and decay time of miniature excitatory postsynaptic currents (mEPSCs) in hippocampal cultures (Fig. 8). USP8 maintains the same number of spines juxtaposed to vGlut1 puncta as control condition (Fig. 7D) and increases (196% ± 25%, P < 0.001, n = 31) the frequency of mEPSCs but does not affect the amplitude and decay time (Fig. 8A and 8B). In contrast, knockdown of USP8 inhibits the leptin-stimulated increase in mEPSC frequency [percent control of mEPSC frequency (Hz) ± SE; control: 100% ± 15%, n = 29; control + leptin: 304% ± 29%, P < 0.001, n = 35; shUSP8: 143% ± 26%, not significant (NS), n = 30; shUSP8 + leptin: 143% ± 19%, NS, n = 29], as well as amplitude [percent control of mEPSC amplitude (−pA) ± SE; control: 100% ± 3%, n = 29; control + leptin: 133% ± 8%, P < 0.001, n = 35; shUSP8: 105% ± 4%, NS, n = 30; shUSP8 + leptin: 106% ± 4%, NS, n = 29; Fig. 8C and 8D]. It should be noted that the modest increase in amplitude seen following leptin stimulation varies from culture to culture (data not shown) for as-yet unknown reasons, whereas the effect on frequency is robust and seen in all cultures. Taken together with the dendritic spine measurements, these data strongly suggest that USP8 increases the formation of functional synapses and is required for leptin-mediated synaptogenesis.

Figure 8.

USP8 regulates synapse formation in hippocampal neurons. (A–D) DIV6 hippocampal neurons were transfected with a fluorescent Clover-β-actin and the indicated constructs. Neurons were stimulated with leptin (50 nM) on DIV8, and electrophysiological recordings were performed on DIV11 to -12. (A and C) Representative electrophysiological recordings. (B and D) Quantification of mEPSC frequency (Hz) normalized to control condition (control: n = 29; V5-USP8: n = 31; control + leptin: n = 35; shUSP8: n = 29; shUSP8 + leptin: n = 30). All experiments were repeated in three independent culture preparations and presented as the means ± SE. ***P < 0.001 compared with control.

Discussion

The CNS actions of leptin primarily require the LepRb, and the size of leptin’s effects are influenced by the number of LepRb (31, 67). However, very little is known about how LepRb protein expression and surface localization are regulated in neurons. Here, we show five findings. First, leptin increases the levels of its own receptor in hippocampal neurons; second, this is through increased deubiquitination of the LepRb receptor, which protects it from targeting for degradation rather than a change in synthesis. Third, leptin induces both the expression and activity of the DUB USP8 in hippocampal neurons. Fourth, USP8 forms a complex with the LepRb and deubiquitinates it, resulting in a reduction in its degradation. Fifth, these actions of USP8 are required for leptin-induced hippocampal synapse formation, indicating that this regulation is critical for one of leptin’s key actions in developing hippocampal neurons.

Leptin enhances LepRb expression

Leptin has been shown to alter LepRb gene expression and protein levels. Early reports from Chinese hamster ovary cell lines show that leptin stimulation decreases LepRb expression (33). However, in vivo peripheral leptin injections increase LepRb mRNA and leptin-binding sites in the hypothalamus, including in the arcuate nucleus (68, 69). Leptin also alters LepRb expression in hypothalamic, pituitary, and ovarian cultures (70). Here, we show that short-term (hours) leptin treatment increases the expression of LepRb protein in both HEK cells and cultured hippocampal neurons. Interestingly, LepRb mRNA levels do not change in hippocampal neurons following leptin stimulation. Changes in LepRb levels have physiological implications, as demonstrated by the partial phenotype of the heterozygous db/db mice (31), suggesting that changes in the levels of LepRb impact the size of leptin’s effects and therefore, the physiological effects mediated by leptin. Interestingly, hypothalamic LepR expression can also be modulated by diet and nutritional state, conditions associated with changes in circulating leptin (71). However, the mechanisms underlying how leptin alters the levels of its own receptor are not well understood. As we found that the effects of leptin are not blocked by inhibiting protein synthesis, our data suggested that the effect was through decreased degradation of the receptor.

Leptin enhances USP8 expression

One mechanism by which the surface LepR expression is controlled is by constitutive endocytosis and degradation via the lysosomal pathway (32, 33, 72). Here, we show that leptin promotes the deubiquitination of LepRb. As ubiquitination targets the LepRb for degradation, deubiquitination of LepRb enhances its expression and promotes the maintenance of LepRb at the plasma membrane of hippocampal neurons.

The DUB USP8 had been shown to regulate LepRb trafficking and degradation in cell lines (34), but its role in the hippocampus had not been determined. Here, we show time that leptin increases USP8 expression levels, which would be expected to increase the level of deubiquitination. Furthermore, USP8 expression and activity are also regulated by post-translational modification (42, 73), and we found that leptin both decreases USP8 pTyr and increases USP8 activity.

USP8 is expressed in other brain regions critical for leptin’s actions, such as the hypothalamus (6–8, 37). It is therefore possible that USP8 deubiquitinates LepRbs in these brain regions, as well increases surface expression of LepRbs and enhances other leptin-dependent behaviors, such as feeding, energy expenditure, and glucose homeostasis.

LepRb is deubiquitinated by USP8

Here, we show that LepRb is deubiquitinated by the DUB USP8 in hippocampal neurons. As it has been shown for many different proteins, deubiquitination of surface proteins prevents their internalization and degradation, whereas deubiquitination is required for proteins to progress from multivesicular bodies into the lysosome to be degraded (74, 75). As enhanced USP8 expression in hippocampal cultures enhances LepRb surface expression, it is most likely that USP8 targets and deubiquitinates surface LepRbs, thus preventing their endocytosis and degradation. We also found that USP8 can form a complex with the LepRb, at least in a cell line, suggesting the enzyme may be in a prime site to deubiquitinate the receptor.

Furthermore, we have demonstrated that USP8 increases LepRb surface localization, as well as promotes activation of two of its key signaling pathways: MAPK kinase/Erk1/2 and CREB. Taken together, our results demonstrate a feed-forward mechanism by which leptin can enhance expression of its own receptor and thus, increase the magnitude of its own signaling.

We also show that leptin stimulates a decrease in ubiquitinated LepRbs, and our data suggest that this is, at least in part, through USP8. However, we cannot rule out that Ub E3 ligases may not also play a role. The E3 ligase RNF41 has been implicated in the recycling of LepRb to the surface (34), but its direct ubiquitination of LepRb or its role in synapse formation has yet to been determined.

Leptin-induced synaptogenesis requires USP8

We have previously shown that leptin stimulates synapse formation in hippocampal neurons (10, 14). Furthermore, this seems to be critical in vivo as well, as dendritic spines are reduced in db/db animals (14). Interestingly, we found that USP8 is required for leptin-induced synaptogenesis. Given that we have shown that leptin stimulates USP8 to increase the expression of its own receptor, the fact that the effects of leptin are lost in the absence of USP8 is most likely a result of enhanced ubiquitination and degradation of LepRb leading to a reduction in LepRb signaling. As LepRbs are required for leptin-induced synaptogenesis, a reduction in LepRb expression would be expected to reduce leptin’s effects.

Our data support previous findings that the increase of USP8 alone is sufficient to promote dendritic spine formation in hippocampal neurons (42), yet we show that there is a reciprocal requirement for LepRbs. The mechanism of this enhancement may include regulation of the expression and activity of AMPA receptors, as USP8 and leptin have both been shown to enhance AMPA receptor activity and surface expression (36, 76). It is therefore possible that leptin enhancement of USP8 expression and activity leads to decreased ubiquitination and enhanced surface expression of AMPA receptors as well. We and others (77) have previously shown that synaptic activity acts as an additional neurotrophic signal to stimulate synaptogenesis and synaptic maturation stability. Thus, it is possible that leptin may produce some of its neurotrophic actions by increasing synaptic activity via the AMPA receptor.

Physiological implication of the contribution of USP8 to the development of synapses in the hippocampus

The Ub signaling system controls a diverse set of neuronal functions, including synapse development and function (38, 39, 78). The removal of Ub from synaptic proteins by DUBs also plays a critical role in synaptic function (40, 41), such as regulation of synaptic activity and structure (79, 80) and regulation of glutamate and GABA-receptor expression (81, 82). Mutations in DUBs have been linked to multiple neurologic disorders, such as Parkinson disease, Angelman syndrome, and ataxia (83–85). The DUB USP8 is highly expressed in multiple brain regions, including the CA1 region and dentate gyrus of the hippocampus, and is also localized in the synapse (35, 36). As multiple synaptic proteins, including Ras protein-specific guanine nucleotide-releasing factor 1, epidermal growth factor receptors, and AMPA receptors, have been shown to be regulated by USP8 deubiquitination (36, 64, 86), USP8 clearly plays an important role in synaptic function.

We show here that USP8 targets LepRb in the hippocampus. Many studies over the past decade have shown that leptin signaling influences hippocampal-dependent behaviors, including having procognitive effects by enhancing learning and memory (87) and having anti-depressive and anxiolytic effects (20, 27). Peripheral and intrahippocampal leptin injections have positive effects on memory tasks, which are dose dependent (25, 88). The actions of leptin in the CNS require the expression of LepRb, which has been observed in the hippocampus (6, 89) with obese db/db mice that lack LepRb expression, showing impairments in spatial learning tasks (90). This correlates well with human studies, as diet-induced obesity has been shown to be associated with decreases in cognitive function, including effects on executive decision making and short-term memory (91, 92). These deficits could arise from leptin insensitivity, both peripherally and centrally, which has been shown to occur in diet-induced obesity (93). Furthermore, leptin signaling has been implicated in multiple neurologic disorders, with decreased levels of LepRb mRNA observed in the hippocampus of patients with Alzheimer’s disease (94), protection of dopamine neurons in a Parkinson disease model (95), and increased levels of circulating leptin presenting in children with autism spectrum disorders (96). As the dendritic spine and synapse number are closely correlated with cognitive function and many neurologic disorders (97–99), furthering our knowledge on the formation of these structures is critical.

In summary, we have identified a signaling pathway by which leptin stimulates dendritic spine and synapse formation in the hippocampus by positively regulating LepRb expression. This occurs through increased expression and activation of USP8, which deubiquitinates LepRb, leading to increased LepRb surface localization and enhanced leptin-stimulated intracellular signaling. Taken together, these results reveal a function of USP8 in synapse formation during early hippocampal development. With the consideration that both LepRb and USP8 are expressed in other brain regions, including the hypothalamus, our findings could have broad implications as to how LepRb expression is regulated, which would, in turn, influence the magnitude of leptin’s effects in these brain regions.

Glossary

Abbreviations:

ACREB

a dominant-negative cAMP response element–binding protein

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor

CA

cornu ammonis

ChIP

chromatin immunoprecipitation

ChIP-Seq

chromatin immunoprecipitation sequencing

CNS

central nervous system

CREB

cAMP response element–binding protein

db/db mice

diabetic mice

DIV

day in vitro

DTT

dithiothreitol

DUB

deubiquitinase

Erk

extracellular signal–regulated kinase

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

HA

hemagglutinin

HEK

human embryonic kidney

IP

immunoprecipitation

K48

lysine 48 residue

LepR

leptin receptor

LepRb

long form of the leptin receptor

LepRb^ΔC

leptin receptor b lacking the intracellular domain

mEPSC

miniature excitatory postsynaptic current

NS

not significant

p

phosphorylated

qRT-PCR

quantitative RT-PCR

shRNA

short hairpin RNA

Tyr

tyrosine

Ub

ubiquitin

Ub^K48R

ubiquitin lysine 48 arginine mutant

Ub^wt

ubiquitin wild type

USP8

ubiquitin-specific protease 8

USP8^C749A

ubiquitin-specific protease 8 cysteine 749 alanine mutant

vGlut1

antivesicular glutamate transporter 1

VME

vinyl methylester

λPP

λ protein phosphatase

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability: Data sharing is not applicable to this article, as no datasets were generated or analyzed during the current study.