Abstract
Leptin is a critical neurotrophic factor for the development of neuronal pathways and synaptogenesis in the hypothalamus. Leptin receptors are also found in other brain regions, including the hippocampus, and a postnatal surge in leptin correlates with a time of rapid growth of dendritic spines and synapses in the hippocampus. Leptin is critical for normal hippocampal dendritic spine formation as db/db mice, which lack normal leptin receptor signaling, have a reduced number of dendritic spines in vivo. Leptin also positively influences hippocampal behaviors, such as cognition, anxiety, and depression, which are critically dependent on dendritic spine number. What is not known are the signaling mechanisms by which leptin initiates spine formation. Here we show leptin induces the formation of dendritic protrusions (thin headless, stubby and mushroom shaped spines), through trafficking and activation of TrpC channels in cultured hippocampal neurons. Leptin-activation of the TrpC current is dose dependent and blocked by targeted knockdown of the leptin receptor. The nonselective TrpC channel inhibitors SKF96365 and 2-APB or targeted knockdown of TrpC1 or 3, but not TrpC5, channels also eliminate the leptin-induced current. Leptin stimulates the phosphorylation of CaMKIγ and β-Pix within 5 min and their activation is required for leptin-induced trafficking of TrpC1 subunits to the membrane. Furthermore, we show that CaMKIγ, CaMKK, β-Pix, Rac1, and TrpC1/3 channels are all required for both the leptin-sensitive current and leptin-induced spine formation. These results elucidate a critical pathway underlying leptin's induction of dendritic morphological changes that initiate spine and excitatory synapse formation.
Introduction
The hormone leptin is a key regulator of energy homeostasis in adults and its effects on food intake and energy regulation have been extensively studied (Ahima and Flier, 2000; Farooqi and O'Rahilly, 2009; Dieguez et al., 2011; Gautron and Elmquist, 2011). However, leptin does not alter food intake or body weight during neonatal developmental period, even though leptin levels surge after birth from postnatal days 7–14 (Ahima et al., 1998). This postnatal surge is essential for the appropriate development of neuronal connections in the hypothalamus, consistent with leptin having neurotrophic actions (Bouret et al., 2004). Leptin receptors (LepR) are expressed widely throughout the brain, including the hippocampus (Huang et al., 1996; Mercer et al., 1996), suggesting that leptin may have neurotrophic actions in other brain regions. Interestingly, the postnatal leptin surge corresponds to a period of rapid development of synaptic connections in the hippocampus. Leptin stimulates synapse formation in the hypothalamus (Pinto et al., 2004) and we have observed that it increases the development of synaptic connections in the hippocampus as well (Dhar et al., 2014).
Deficiency in either leptin or LepR signaling leads to impaired hippocampal function and associated aberrations in hippocampal behaviors, such as depression, anxiety, and decreased memory (Li et al., 2002; Sharma et al., 2010; Yamada et al., 2011). Leptin has also been shown to modulate hippocampal plasticity, such as long-term potentiation (LTP) or depression (Beccano-Kelly and Harvey, 2012). The long-form LepR (LepRb) deficient mice (db/db) have lowered spine density in the dentate gyrus, CA1, and CA3 regions of the hippocampus compared with their wild-type counterparts (Stranahan et al., 2009; Dhar et al., 2014). However, although leptin has been shown to stimulate formation of protrusions (filopodia; O'Malley et al., 2007), the first stage to forming a spine, the molecular mechanisms by which leptin initiates this critical process are unknown.
Leptin has recently been shown to activate a TrpC current in the hypothalamus (Qiu et al., 2010; Williams et al., 2011) and TrpC channels are important for brain derived neurotrophic factor (BDNF)-induced spine formation in the hippocampus (Amaral and Pozzo-Miller, 2007a,b; Li et al., 2010). Therefore, in this study we determined whether leptin activates a TrpC current in developing hippocampal neurons and whether this current is required for leptin-induced spine formation. As TrpC channels are rapidly inserted into the membrane by other growth factors (Bezzerides et al., 2004; Amaral and Pozzo-Miller, 2007b), we also determined whether leptin stimulated trafficking of TrpC channels. Last, we determined whether Rac1 was required and whether it involved a novel signal transduction pathway for leptin, the calcium/CaM-dependent kinase (CaMK) cascade, as this cascade has been shown to be an important regulator of the actin cytoskeleton and synapse formation (Nakayama et al., 2000; Saneyoshi et al., 2008; Wayman et al., 2008).
Materials and Methods
Drugs and DNA constructs.
Physiologically active synthetic leptin peptide fragment (116-130) was purchased from Tocris Bioscience and used at 50 nm concentration. STO-609 and SKF96365 were purchased from Tocris Bioscience and used at 20 and 3 μm concentration respectively. All other drugs were purchased from either Sigma-Aldrich or Tocris Bioscience. shLepRb, GCTCACTGTCTGTTCAGTGAC was cloned in the pSUPER vector using OligoEngine protocol. shTrpC1, CCTTGAAGATCGTGGCCTATG and shTrpC5, CCTTGAAGTCGTGGCCTATG were cloned in the pU6 siRNA vector under the SV40 promoter. ShRNA constructs for CaMKKα, CaMKKβ, CaMKIβ, CaMKIγ, and β-Pix were previously described (Wayman et al., 2006; Saneyoshi et al., 2008). Mutants for β-Pix (S516A and DHm) were described by Saneyoshi et al. (2008). Rac1 constructs (shRNA and dominant-negative) and dominant-negative Pak1 construct was described by Impey et al. (2010).
Hippocampal cell culture preparation.
Hippocampal neurons (3 × 104 cells/cm2 for 24-well plates used for spine and electrophysiology experiments and 4.7 × 104 cells/cm2 for 6-well plates used for biochemistry experiments) were cultured from P1 Sprague-Dawley rats on plates coated with poly-l-lysine from Sigma-Aldrich (molecular weight 300,000). At this age, it is extremely difficult to reliably determine the sex of the pups and so we included both male and female pups for our cultures. Hippocampal neurons were maintained in neurobasal A medium from Invitrogen supplemented with B27 (Invitrogen) and 0.5 mm l-glutamine and 5 mm cytosine-d-arabinofuranoside (Sigma-Aldrich) added at 2 d in vitro. Hippocampal neurons were then cultured a further 3–7 d, at which time they were either transfected or treated with various pharmacological reagents as described by Wayman et al. (2008).
Whole-cell recordings.
Patch-clamp experiments were performed on mRFP-β-actin-transfected cultured DIV8 to DIV9 hippocampal neurons. Transfected cells were visualized with fluorescence (IX-71, Olympus Optical). Recordings were made at room temperature with the membrane potential held at −70 mV in an extracellular solution of 140 mm NaCl, 2.5 mm KCl, 1 mm MgCl2, 3 mm CaCl2, 25 mm glucose, and 5 mm HEPES, pH 7.3, 305–310 mOsM, and with 100 μm picrotoxin, 1 μm strychnine, and 500 nm tetrodotoxin included in the external solution. The resistance of patch electrodes ranged from 4.0 to 5.2 MΩ, with an internal solution of: 25 mm CsCl, 100 mm CsCH3O3S, 10 mm phosphocreatine, 0.4 mm EGTA, 10 mm HEPES, 2 mm MgCl2, 0.4 mm Mg-ATP, and 0.04 mm Na-GTP, pH 7.2, 296–300 mOsM. Recordings were obtained using a Multiclamp 700B amplifier (Molecular Devices), Bessel-filtered at 2 kHz, digitized at 10 kHz through a Digidata 1440A interface (Molecular Devices), and acquired using Clampex 10.2 software (Molecular Devices). Only neurons with a >2 GΩ seal and input resistance >240 MΩ were used and liquid junction potentials were not corrected. Data analysis was performed using Clampfit 10.2 software (Molecular Devices). Reported currents were measured from baseline to peak. On average, it took 60 s for the perfusion solution to fill the chamber. This “dead time” was subtracted from the onset time of the reported current.
Current–voltage (I–V) ramps were performed as described by Qiu et al., 2010. Briefly, the external solution contained the following (in mm): CsCl 128.5, CaCl2 1.4, MgCl2 1.2, HEPES 20.0, d-glucose 10, CsCl 2.0, CoCl2 1.0, 4-aminopyridine 5, picrotoxin 0.1, nifidipine 0.01, TTX 0.001, NaOH 8, pH7.3, 308–310 mOsM. The resistance of patch electrodes ranged from 3.7 to 4.7 MΩ, with an internal solution of (in mm): Cs-gluconate 125, NaCl 10, MgCl2 1, EGTA 11, HEPES 10, Mg-ATP 4, and Na-GTP 0.25, pH 7.3, by CsOH, 298–300 mOsM. I–V ramps were performed from −100 to + 40 mV from a holding potential of −60 mV in the presence and absence of 100 nm leptin.
Transfection.
For immunoprecipitation, neurons were transfected with the desired constructs, such as βPix-myc for βPix phosphorylation experiments and TrpC1-flag for biotinylation experiments, along with various shRNAs or dominant-negative constructs on DIV5–6 using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. This protocol yielded a low 3–5% transfection efficiency, but the protein was concentrated using appropriate immunoprecipitation protocols mentioned below.
For spine analysis, neurons were transfected with mRFP-βactin along with various other DNA constructs, and shRNAs on DIV 6 using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. This protocol yielded the desired 3–5% transfection efficiency thus enabling the visualization of individual neurons. Expression of fluorescently tagged actin allowed clear visualization of transfected neurons and their dendritic spines, because dendritic spines are enriched in actin.
Protrusion quantification.
On DIV8, cells were treated with 50 nm leptin (as described in Results) added to media. In the case of SKF or STO-609 pretreatment, cultures were incubated with the required drug added to the media for 30 min before leptin stimulation. Immediately after a 30 min leptin treatment, the neurons were fixed (4% paraformaldehyde in PHEMS buffer, pH 7.4) for 20 min at room temperature and mounted on glass slides using elvanol. Slides were cured overnight at 4°C, and fluorescent images were obtained with Slidebook 5.0 digital microscopy software driving an inverted Olympus IX81-DSU microscope) with a 60× oil-immersion lens, numerical aperture 1.42, and a pixel dimension of 0.108 μm. Protrusion density was measured on primary and secondary dendrites at a distance of at least 100 μm from the soma. Two to five dendrites, each at least 50 μm in length, from at least 22 neurons were analyzed for each data point reported. Each experiment was repeated at least three times using independent culture preparations. Dendrite length was determined using ImageJ 1.41o (NIH) and the neurite tracing program Neuron J (Meijering et al., 2004).
As we are studying the initiation of spine formation we chose to use younger neurons in culture (days 8–10). Protrusions were manually counted and the data in the figures is presented as number of protrusions per 50 μm of dendritic length. Protrusions were further classified into three categories: mature (stubby and mushroom) and thin headless spines, and these data are presented in Table 2. Thin headless spines were identified as short dendritic protrusions with a long, thin neck and without an obvious head. Mature spines were identified as small protrusions (≤3 μm) with actin-enriched spherical head connected to a dendrite via a thin neck (Segal, 2005; Bourne and Harris, 2008; Svitkina et al., 2010). The head diameter is at least twice the diameter of the neck. These mature spines were further characterized as either stubby or mushroom-type depending on the neck length. Stubby spines have a short or nonexistent neck, whereas mushroom spines have a more discrete neck (Segal, 2005; Bourne and Harris, 2008).
Immunocytochemistry.
For shRNA validation, neurons were transfected with pCAGGS-tdTomato (a red fluorescent protein to identify transfected neurons), a plasmid expressing an epitope-tagged target protein along with shRNA to be tested on DIV5–6. After fixation on DIV8, cells were rinsed in PBS and permeabilized with 0.1% Triton X-100 detergent (Bio-Rad Laboratories), followed by two rinses in PBS, and blocked with 0.5% fish gelatin in PBS for 2 h. Cells were rinsed with PBS again, followed by a 24 h incubation period with anti-flag (Sigma-Aldrich, for TrpC1-flag), or anti-HA (Sigma-Aldrich, for TrpC5-HA), following the manufacturer's protocol, at 4°C. Then, cells were rinsed twice with PBS, incubated in AlexaFluor 488 secondary IgG antibody (Invitrogen) following the manufacturer's protocol for 2 h at room temperature, rinsed again with PBS, and mounted with elvanol. Imaging for staining was performed using Slidebook 5.0 digital microscopy software driving an inverted Olympus IX81-DSU microscope with a 20× lens, numerical aperture 0.5, and a pixel dimension of 0.33 μm. Using the Slidebook 5.0 digital microscopy software intensity expression of the tagged construct was measured in transfected neurons by creating a mask in the red channel (identifies transfected neurons) and measuring mean signal intensity in the green channel (identifies tagged plasmid expression levels).
Western blot.
Protein samples were collected from hippocampal cultures plated at ∼4.7 × 104 cells/cm2. Post-treatment cells were lysed with RIPA buffer (Sigma-Aldrich) augmented with Phosphatase inhibitor cocktail 2 and 3 (Sigma-Aldrich), followed by centrifugation at 14,000 rpm for 15 min to collect the protein-enriched supernatant. The supernatant collected was mixed with SDS-loading buffer (Invitrogen) and DTT, boiled for 5 min, and loaded on a SDS-PAGE (Invitrogen). Western blotting was done using antibodies at the following dilution: anti-CaMK1 (gift from Monika Davare, Oregon Health and Sciences University, Portland, OR), pCREB (Millipore Bioscience Research Reagents, 1:1000), anti-myc (Sigma-Aldrich, 1:5000), anti-Flag (Sigma-Aldrich, 1:1000), anti-phosphoβPix (gift from T. Saneyoshi, Brain Science Institute, RIKEN, Wako, Saitama, Japan; 1:200), avidin-IRDye700 (Rockland, 1:10,000). Blots were imaged using Odyssey infrared detection system and analyzed using the ImageJ (NIH) gel analyzer tool.
Immunoprecipitation.
For immunoprecipitation, protein samples were collected as described above and proteins of interest immunoprecipitated using the corresponding antibody (anti-myc antibody for βPix IP, anti-flag M2 antibody for TrpC1 IP) for 18 h at 4°C. The antibody-protein complex was pulled down using protein-A agarose beads (Millipore) for 4 h at 4°C. Collected beads were washed with lysis buffer and eluted in 2× SDS/loading buffer (Invitrogen) at 50°C for 30 min. Eluted protein was sonicated and boiled for 5 min before loading on a SDS-PAGE.
Biotinylation.
A TrpC1 construct tagged with a flag epitope was transfected in DIV5 hippocampal neurons and on DIV8 neurons were treated with 50 nm leptin. After leptin stimulation, hippocampal cultures were first washed with ACSF for 20 min and then treated with 1 mg/ml NHS-LC-Biotin (Pierce) for 30 min. Finally, the cells were washed with a Quenching ACSF for 10 min. After the biotinylation steps, cells were lysed as described previously and immunoprecipitated for TrpC1-flag using anti-flag M2 antibody (Sigma-Aldrich). After accounting for transfection and pull-down efficiency, all samples were compared for membrane-bound TrpC1 levels using an avidin antibody that specifically binds biotinylated proteins.
Statistical analysis.
Differences in drug effects in electrophysiology experiments were tested either by a repeated measured or one-way ANOVA followed by a Tukey's post hoc test. One-way ANOVA was used to analyze the protrusion results and significant effects were analyzed by Tukey post hoc test. Sample size for each reported condition in the protrusion measurement experiments from hippocampal cultures is at least 25 neurons collected from three independent experiments. Western blots were analyzed using two-tailed unpaired Student's t test. Sample size for each reported condition from hippocampal cultures analyzed using Western blots was two to three biological replicates collected from two to three independent experiments. Multiple numerical data are expressed as means ± SEM. Statistical information is given in the text for any data not included in the figures. Otherwise, all statistical information is provided in the figures and figure legends. A summary of all the statistics performed for the protrusion and spine data counted for each condition are shown in Tables 1 and 2. Table 1 also shows the total number of neurons, number of dendrites, average dendritic length, and total dendritic length analyzed for each condition.
The number of neurons, dendrites, and total dendritic length analyzed for each condition
The effect of all treatment conditions on stubby, mushroom, and thin headless spines
Results
Acute leptin treatment induces protrusion formation as well as an inward current in hippocampal neurons
Dendritic protrusions can be either headless spines, which are potential sites of future excitatory synaptic connections on a neuron, or mushroom or stubby spines, which are more mature spines that have a prominent head and are normally associated with functional glutamatergic synapses (Segal, 2005; Bourne and Harris, 2008). Neurons in vitro develop in a defined pattern: starting from axonal specification, then dendritic arborization, and finally spine and synapse formation (Bartlett and Banker, 1984a,b). Our cultures of DIV8–10 neurons are in a phase of rapid synapse development and we found that 45% of the protrusions were headless spines (with 92% being <3 μm in length) and 55% were spines with prominent heads (head diameter of at least twice the size of neck diameter) including either mushroom (17%) or stubby (38%) types. Various neurotrophic signals can create permissive conditions that prompt a neuron into developing more synaptic contacts with its neighboring neurons. Similar to previous reports, we observed a rapid increase in total dendritic protrusions in DIV8 hippocampal neurons after 30 min of leptin stimulation (Figure 1A,B; Table 1; O'Malley et al., 2007). Leptin treatment increases the density of all spine types compared with vehicle-treated controls. It increased headless spine density from 6.2 ± 0.2 to 9.4 ± 0.4, stubby spines from 5.3 ± 0.2 to 8.5 ± 0.3 and mushroom-type spines from 2.4 ± 0.1 to 4.7 ± 0.3 (numbers are spines per 50 μm ± SEM; Table 2). All spine-types are quantified together as dendritic protrusions in the figures, but the effects of leptin on each spine type are shown in the tables for the rest of the paper. Leptin's induction of protrusion and spine formation was mediated by the long isoform of its receptor (LepRb) because neurons transfected with a targeted shRNA to knockdown LepRb expression did not show any leptin-induced increase in protrusion density (Fig. 1A,B; Control = 14.0 ± 0.4; leptin = 22.6 ± 0.5; shLepR = 18.8 ± 1.2, shLepR + leptin = 17.4 ± 1.8; Fig. 1; Tables 1, 2). The shLepRb also attenuated leptin-induced increases in stubby and mushroom spines (Table 2). Interestingly, shLepRb alone significantly increased the number of headless spines, suggesting that a functional leptin receptor may be required for maturation of the headless spines into mushroom or stubby spines; there was also a trend toward reduction in both these types of mature spines. In another complimentary study we have observed that in older cultures (DIV 12–14) shLepR significantly reduces stubby and mushroom spines (Dhar et al., 2014). Leptin also increased the average spine head size, an effect blocked by shLepR (Control = 0.62 ± 0.02 μm, n = 45; leptin = 0.82 ± 0.03 μm, n = 44; shLepR = 0.68 ± 0.01 μm, n = 44; shLepR+leptin = 0.59 ± 0.02 μm, n = 40; leptin vs Control: p < 0.005, leptin vs shLepR + leptin: p < 0.005, Control vs shLepR + leptin: NS; one-way ANOVA), again suggesting that leptin promotes the formation of more mature spine types.
Leptin induces actin-enriched dendritic protrusions as well as an inward current in hippocampal neurons. A, B, Cultured hippocampal neurons were transfected with mRFP-βactin ±shLepR on DIV6 and treated with vehicle or leptin on DIV8 for 30 min. A, Representative images; B, graph showing average protrusion density under specified conditions. C, Indicated concentration of leptin was bath applied onto DIV8–9 hippocampal neurons and the resulting current was recorded. Graph shows average peak amplitude at specified concentrations. D, I–V relationship for the leptin-activated current. The graph shows the subtracted leptin-activated current averaged from five neurons (I–V ramp from control conditions subtracted from the I–V ramp obtained in the presence of leptin). E, F, Cultured hippocampal neurons were transfected with mRFP-βaction ±shLepR on DIV6 and leptin current was recorded on DIV8–9. D, Representative traces; E, graph showing average peak amplitude of leptin current under control and shLepR conditions. Protrusion and electrophysiology data were analyzed using ANOVA followed by Tukey's post hoc analysis (±SEM; for protrusions ***p < 0.001 compared with control, vehicle, **p < 0.01 compared with leptin; for Ileptin ***p < 0.001 compared with the leptin-induced current under control conditions).
LepRs have been shown to be present in hippocampal neurons by both mRNA and protein expression (Huang et al., 1996; Shioda et al., 1998; Shanley et al., 2002;Scott et al., 2009) suggesting that leptin could have direct effects on hippocampal neurons. Leptin also activates both pCREB and pErk in the hippocampus, although unlike other brain regions it does not always increase pSTAT3 (Walker et al., 2007; Zhang and Chen, 2008; Caron et al., 2010; Dhar et al., 2014). Furthermore, leptin has been shown to have effects on hippocampal neurons as measured by electrophysiological techniques, which are seen in isolated hippocampal slices and therefore must be locally mediated (for review, see Beccano-Kelly and Harvey, 2012). We therefore determined whether leptin has direct effects in hippocampal neurons that could mediate its effects on spine induction by measuring currents before and after leptin application. Leptin stimulated a slow activating inward current (Ileptin) in DIV8–9 hippocampal cultured neurons (Figure 1C,D). Leptin-activation of the current was concentration-dependent (Figure 1C; leptin current, pA: 1 nm = −0.9 ± 1.8, n = 3; 3 nm = −1.5 ± 2.3, n = 5; 10 nm = −7.9 ± 2.2, n = 6; 30 nm = −18.7 ± 1.9, n = 3; 50 nm = −19.0 ± 1.6, n = 8; 100 nm = −21.3 ± 2.0, n = 7) and the effect was mediated by LepRb, as reducing LepRb expression by shRNAs completely blocks Ileptin (Fig. 1E,F; leptin current, pA: Control = −30.5 ± 1.5, n = 9; shLepR = −0.2 ± 1.0, n = 16; p < 0.001 compared with Ileptin under control conditions by one-way ANOVA). The average onset time was 4.1 ± 0.2 min; the average rise time was 4.3 ± 0.1 min and the average time to peak was 8.4 ± 0.2 min. We also performed an I–V ramp to characterize Ileptin (Fig. 1D). The I–V relationship of Ileptin is remarkably similar to that reported for the TrpC current activated by leptin in POMC neurons in the hypothalamus (Qiu et al., 2010). The I–V relationship is also similar to those reported following heterologous expression of TrpC1 as a heteromultimer with TrpC3, 4, or 5 (Strübing et al., 2001, 2003; Storch et al., 2012).
Pharmacological characterization of leptin-induced current
As leptin activates TrpC-mediated currents in the hypothalamus (Qiu et al., 2010), we next determined whether this slow-activating current was mediated by TrpC channels. Pretreatment with SKF96365 (SKF), a general TrpC channel blocker (Clapham et al., 2005), attenuated both leptin's effects on protrusion formation (Fig. 2A,B; SKF = 14.1 ± 0.8, SKF + leptin = 15.4 ± 1.0) and the leptin-induced current (Fig. 2C,D; leptin current, pA: Control = −20.4 ± 1.1, n = 5; with SKF = −0.5 ± 2.7, n = 5; p < 0.001 compared with Ileptin under control conditions by one-way ANOVA). Extracellular application of another TrpC inhibitor, 2-APB, (Clapham et al., 2001), also attenuated Ileptin supporting the hypothesis that the leptin-activated current required functional TrpC channels (Fig. 2D; 2-APB = −1.9 ± 2.3 pA, n = 5; p < 0.001 compared with Ileptin under control conditions by one-way ANOVA). 2-APB also inhibits IP3 receptors and voltage-gated calcium channels (Bootman et al., 2002) and SKF is also relatively nonselective as it also inhibits voltage-gated calcium channels (Merritt et al., 1990; Zhao and Simasko, 2010). For this reason, we next investigated the role of TrpC channels further using other pharmacological approaches and by specific knockdown with shRNAs.
Leptin-induced protrusion formation and current are blocked by SKF. A, B, Cultured hippocampal neurons were transfected with mRFP-βactin on DIV6 and treated with leptin on DIV8 for 30 min. SKF (3 μm) treatment was applied 20 min before leptin treatment. A, Representative images; B, graph showing average protrusion density under specified conditions. C, D, Leptin was bath applied to DIV8–9 cultured hippocampal neurons and leptin current was recorded. Prior exposure to SKF blocked the leptin response, which was restored following wash out of SKF. C, Representative traces; D, graph showing average peak amplitude of leptin current under control and SKF conditions. Protrusion and electrophysiology data were analyzed using ANOVA Tukey's post hoc analysis (±SEM; for protrusions: ***p < 0.001 compared with control, vehicle, ***p < 0.001 compared with leptin; for Ileptin ***p < 0.001 compared with the leptin-induced current under control conditions).
Leptin current requires TrpC 1 and 3, but not 5 expression
The TrpC channel family consists of seven different subunits. Most of which can form either homomeric or heteromeric channels (Clapham et al., 2005). The presence of specific subunits in a TrpC heteromer determines its pharmacological properties and ionic permeability. At millimolar concentrations, lanthanum is a general blocker of TrpC channels but at micromolar concentrations it potentiates TrpC4 or 5 channel currents (Schaefer et al., 2000; Strübing et al., 2001). In contrast to the leptin current observed in the arcuate nucleus (Qiu et al., 2010), 100 μM La3+ did not potentiate Ileptin in hippocampal neurons. At 100 μm, lanthanum partially reduced, and at 1 mm, lanthanum completely blocked Ileptin (leptin current, pA: Control = −20.5 ± 1.1, n = 32; 100 μm La3+ = −12.8 ± 3.4, n = 6; 1 mm La3+ = −0.6 ± 1.1, n = 9; p < 0.01 and p < 0.001, respectively, compared with Ileptin under control conditions by one-way ANOVA; Fig. 3A), suggesting that the TrpC channel mediating Ileptin does not contain TrpC4 and 5 subunits.
Leptin current is mediated by TrpC1 and 3 channel. A, DIV8–9 cultured hippocampal neurons were treated as specified and leptin current was recorded. Amplitude of leptin current under indicated conditions is shown. B, C, Dissociated hippocampal cultures were transfected with dsRED to identify transfected neurons and TrpC1-flag or TrpC5-HA construct alone or along with respective shRNA. Cultures were fixed and immunostained for TrpC1 expression using anti-flag antibody and TrpC5 expression using anti-HA antibody. D, E, Cultured hippocampal neurons were transfected with mRFP-βactin ± shTrpC1, siTrpC3, shTrpC5, or scrambled shRNA on DIV6. Leptin was bath applied on DIV8–9 neurons and leptin current was recorded. D, Representative traces; E, graph showing average peak amplitude of leptin current under indicated conditions. Data were analyzed using ANOVA Tukey's post hoc analysis (±SEM; **p < 0.01, ***p < 0.001 compared with the Ileptin, leptin-induced current, under control conditions).
TrpC channels are generally considered nonselective cationic channel that are permeable to both Ca2+ as well as Na+ with a subunit-dependent preference (Clapham et al., 2001). Reducing the calcium concentration from 3 to 0.5 mm in the bath had no significant effect on the amplitude of Ileptin, whereas reducing the sodium concentration from 140 to 5 mm strongly reduced the amplitude of the leptin-induced current (Fig. 3A). Furthermore, we found Ileptin was partially reduced by increasing extracellular magnesium (leptin current, pA: Control 140 mm sodium, 2 mm Mg2+ = −20.7 ± 1.2, n = 79; 5 mm sodium, 2 mm Mg2+ = −2.1 ± 2.6, n = 4; 0.5 mm Ca2+, 0 mm Mg2+= −18.0 ± 1.4, n = 6; 0 mm Ca2+, 4 mm Mg2+ = −6.7 ± 2.0, n = 6; 5 mm Na+, 2 mm Mg2+ = p < 0.001 and 0 mm Ca2+, 4 mm Mg2+ = p < 0.01 compared with Ileptin under control conditions by one-way ANOVA; Fig. 3A). These properties are again consistent with a TrpC1/3/6 type of channel (Clapham et al., 2005).
To confirm the molecular composition of the TrpC channel involved we used shRNAs or siRNAs to reduce expression of specific TrpC subunits and determine their effect(s) on the leptin-induced current. The abilities of the shRNAs used to target TrpC1 and TrpC5 to knockdown protein expression were validated in neurons using immunostaining (Fig. 3B,C). The validity of the siRNA targeting TrpC3 has been previously confirmed in (Amaral and Pozzo-Miller, 2007b). Expression of TrpC1 and TrpC3 was found to be essential for the leptin-induced current as targeted knockdown of either subunit blocked Ileptin (Fig. 3D,E; leptin current, pA: Control = −23.7 ± 1.9, n = 12; shTrpC1 = −0.7 ± 1.2, n = 9; siTrpC3 = 0.4 ± 1.5, n = 9; scrambled shRNA = −21.7 ± 1.3, n = 3; shTrpC1 and siTrpC3 = p < 0.001 compared with Ileptin under control conditions by one-way ANOVA). However, consistent with our pharmacological data, TrpC5 expression was not required as shRNAs targeting TrpC5 had no effect (Fig. 3D,E; leptin current with shTrpC5 = −17.5 ± 1.5 pA; n = 7; p > 0.05 compared with Ileptin under control conditions by one-way ANOVA).
Leptin initiates the CaMK pathway to activate TrpC current
We next determined the molecular pathway induced by leptin that activates TrpC1/3 channels. Leptin increased CaMKI phosphorylation/activation in DIV8 cultured hippocampal neurons within 5 min, an effect that lasted for >10 min (Fig. 4A,B; 5 min = p < 0.05, 10 min = p < 0.01, Student's t test). Consistent with this pathway being required for activation of Ileptin, pretreatment with STO-609, a CaMKK inhibitor, completely attenuated leptin's induction of the current (Fig. 4C; Current, pA: leptin = −21.3 ± 2.1, n = 7; STO = −0.4 ± 1.0, n = 10; STO + leptin = −0.4 ± 1.7, n = 10; p < 0.01 compared with Ileptin under control conditions by one-way ANOVA). Moreover, targeted knockdown of specific isoforms of CaMKK also attenuated Ileptin. Transfection of shCaMKKβ completely blocked Ileptin, whereas shCaMKKα partially reduced the leptin current (Fig. 4C). We next determined that CaMKIγ was required for Ileptin as shCaMKIγ, but not shCaMKIβ, completely attenuated the current (Fig. 4C; leptin current, pA: Control = −21.6 ± 0.9, n = 52; shCaMKKα = −12.9 ± 1.6, n = 7; shCaMKKβ = 0.5 ± 1.0, n = 6; shCaMK1γ = −1.9 ± 2.4, n = 7; shCaMK1β = −16.6 ± 1.0, n = 6; p < 0.001 compared with Ileptin under control conditions by one-way ANOVA). Furthermore, intracellular calcium is required for induction of Ileptin as leptin's effects were blocked by intracellular BAPTA (Fig. 4C; 25 mm BAPTA + leptin = 1.2 ± 1.2 pA, n = 7; p < 0.001 compared with Ileptin under control conditions by one-way ANOVA). Calcium has also been shown to be required for normal spine formation (for review, see Wayman et al., 2011).
Leptin-activated CaMK pathway is required for the leptin current. A, B, DIV8 cultured hippocampal neurons were treated with leptin for various durations (5–30 min) and then lysed using RIPA. Samples were run on a SDS-PAGE and blotted with anti-pCaMKI and anti-Erk1/2 antibody. A, Representative Western blot; B, average pCaMKI band intensity normalized to Erk1/2 (loading control) under indicated conditions is shown. C, Cultured hippocampal neurons were transfected with mRFP-βactin ± shCaMKKα, shCaMKKβ, shCaMKIγ, or shCaMKIβ on DIV6. Leptin was bath applied on DIV8–9 neurons and leptin current was recorded. STO-609 (20 μm) and BAPTA (25 mm) was bath applied before leptin application. Graph showing average peak amplitude of leptin current under indicated conditions. Western blot data were analyzed using Student's t test (±SEM, *p < 0.05, **p < 0.01 compared with untreated). Electrophysiological data were analyzed using ANOVA Tukey's post hoc analysis (±SEM; *p < 0.05, **p < 0.01, ***p < 0.001 compared with Ileptin under control conditions).
β-Pix-activated Rac signaling is required for leptin-induced TrpC current
We have previously shown that CaMKK, CaMKI, and β-Pix, a Rac guanine exchange factor (GEF), form a signaling complex in hippocampal neurons and activate Rac1 and stimulate synaptogenesis in response to synaptic activity (Saneyoshi et al., 2008). Therefore, we next determined whether leptin increased the phosphorylation state of β-Pix and whether this phosphorylation event is CaMKK/CaMKIγ-dependent. To test this, hippocampal neurons were transfected with myc-tagged β-Pix alone or along with short-hairpin constructs targeting the CaMK pathway. β-Pix was then immunoprecipitated with an anti-myc antibody following leptin stimulation. Similar transfection and pull-down efficiency across samples was confirmed by comparing input samples as well as immunoprecipitated samples using an anti-myc antibody (Figure 5A). Leptin treatment increases β-Pix phosphorylation 5–10 min post-treatment (Fig. 5A,B; p < 0.05 Student's t test). The leptin-induced phosphorylation of β-Pix was dependent on CaMKK/CaMKIγ as leptin had no effect in neurons pretreated with STO-609 or transfected with shCaMKIγ (data not shown; Fig. 5A,B). We next confirmed the requirement of β-Pix and Rac1 activity for Ileptin. Blocking β-Pix and Rac1 expression by shRNAs (shβ-Pix and shRac1) or activity by dominant negatives (dnRac1 for Rac1) blocked Ileptin, as did overexpression of DHm, a β-Pix mutant lacking the GEF domain or β-PixS516A, a mutant construct that cannot be phosphorylated by CaMKI (Fig. 5C; leptin current, pA: Control = −21.6 ± 0.9, n = 52, same as above; shβ-Pix = −1.1 ± 0.9, n = 9; β-PixS516A = −0.2 ± 1.8, n = 14; dnRac1 = 0.7 ± 1.6, n = 11; dnPak1 = −1.1 ± 0.8, n = 9; shRac1 = 0.9 ± 1.7, n = 10; DHM = 1.1 ± 1.8, n = 7; p < 0.001 for all treatments compared with Ileptin under control conditions by one-way ANOVA).
β-Pix phosphorylation and activation is required for leptin current. A, B, DIV5 cultured hippocampal neurons were transfected with β-Pix-6myc ±shCaMKIγ and stimulated with leptin for various durations on DIV8. Cells were lysed in RIPA and immunoprecipitated using anti-myc antibody. Samples were loaded on a SDS-PAGE and blotted with anti-phospho β-Pix and anti-Myc antibody. A, Representative Western blot; B, graph showing average relative phospho β-Pix band intensity of immunoprecipitated samples normalized to input samples for indicated conditions. C, DIV5 cultured hippocampal neurons were transfected with mRFP-βactin along with the specified constructs. Leptin was bath applied to DIV8–9 neurons and leptin current was recorded. Graph shows the average peak amplitude of leptin current under indicated conditions. Western blot data were analyzed using Student's t test (±SEM, *p < 0.05, compared with control, untreated). Electrophysiological data were analyzed using ANOVA Tukey's post hoc analysis (±SEM; ***p < 0.001 compared with Ileptin under control conditions).
Leptin increases TrpC1 trafficking to the membrane via a CaMK/βPix-dependent pathway
The slow-initiation of the leptin-activated TrpC current in our experiments could be attributed to a required trafficking event. To test this hypothesis, we conducted surface biotinylation experiments to measure changes in the amount of TrpC channels in the neuronal membrane before and after leptin stimulation. Leptin increased TrpC1 levels in the plasma membrane 10 min after leptin treatment and this increase was sustained for >20 min (Fig. 6A,B; p < 0.01, Student's t test). Leptin did not alter total TrpC1 levels, as there was no difference in the input samples between conditions. Cotransfection of the TrpC1-flag construct with shCaMKIγ or β-Pix(S516A) did not change the basal levels of TrpC channel in the membrane compared with controls but they completely blocked leptin from stimulating TrpC trafficking (Fig. 6A,B).
Leptin increases TrpC1 trafficking to the membrane via a CaMK/βPix-dependent pathway. A, B, DIV5 cultured hippocampal neurons were transfected with TrpC1-flag ±shCaMKIγ or βPixS516A and treated with leptin on DIV8 for various durations. Cells were biotinylated, quenched and lysed, and immunoprecipitated using anti-flag antibody. A, Representative Western blot; B, relative average intensity of avidin band normalized to flag band under indicated conditions. Western blot data were analyzed using Student's t test (±SEM; **p < 0.01 compared with control, untreated).
TrpC channel trafficking and activity is required for leptin-induced spine initiation in hippocampal neurons
Last, we determined the requirement of the signaling cascade initiated by leptin for TrpC channel trafficking in leptin's effects on protrusion formation. Inhibiting the CaMK cascade by blocking CaMKK activity using STO-609 or CaMKIγ expression using a shRNA blocks leptin's effects on protrusion formation (Fig. 7A,B; Control = 14.0 ± 0.4, leptin = 22.6 ± 0.5, STO = 16.5 ± 1.0, STO + leptin = 16.7 ± 1.0, shCaMKIγ = 14.9 ± 1.0, shCaMKIγ + leptin = 15.5 ± 1.1; Tables 1, 2 show detailed spine analysis and statistical comparisons). Activation of the β-Pix phosphorylation and subsequently Rac1 activation is also required as leptin also failed to induce protrusions in neurons transfected with β-PixS516A or shRac1 (Fig. 7A,B; β-PixS516A = 16.0 ± 0.9, β-PixS516A + leptin = 14.4 ± 1.0, shRac1 = 13.4 ± 1.1, shRac1 + leptin = 13.0 ± 0.9, p < 0.001 one-way ANOVA). TrpC1 and 3 were also required for leptin-induced protrusion formation, consistent with them forming the required TrpC channel as targeted knockdown of either blocked leptin's effects as did application of SKF (Fig. 7A,C; shTrpC1 = 14.6 ± 0.8, shTrpC1 + leptin = 15.8 ± 0.9, siTrpC3 = 13.3 ± 1.1, siTrpC3 + leptin = 12.1 ± 0.8, p < 0.001 one-way ANOVA). Together, these data confirm that the signaling cascade that we describe for leptin-induced TrpC trafficking is also required for leptin-induced protrusion and spine formation. The effects of all the treatments on the individual spine types (mushroom, stubby, and headless) are shown in Table 2. Inhibition of TrpC channels with SKF, shTrpC1, or shTrpC3 all blocked leptin-induced formation of stubby and mushroom spines (Table 2). Furthermore, inhibition or knockdown of the CaMKK/CaMKI pathway and of β-Pix and Rac1 also all blocked leptin-induced formation of different types of spines (Table 2). Interestingly, both SKF and shTrpC1 also caused an increase in control (basal) thin, headless spine formation similar to shLepR, which was again associated with a trend to reduce stubby and mushroom type spines.
TrpC channel trafficking and activity is required for leptin-induced spine initiation in hippocampal neurons. A–C, DIV5 cultured hippocampal neurons were transfected with mRFP-βactin along with various specified constructs. Neurons were stimulated with leptin for 30 min on DIV8 and then fixed immediately. A, Representative images; B, C, average protrusion density under indicated conditions. Data were analyzed using ANOVA followed by Tukey post hoc analysis [±SEM; ***p < 0.001 compared with control (vehicle) or leptin].
Discussion
Leptin induces synapse formation in both the hypothalamus and hippocampus (Pinto et al., 2004), and this is thought to be a key mechanism by which leptin modulates the activity and strength of neuronal pathways and circuits. Furthermore, db/db mice lacking functional leptin receptors have fewer hippocampal dendritic spines than wild-type controls (Stranahan et al., 2009), suggesting that leptin is required for normal development of spines and synaptic connections in the hippocampus. What was not known were the critical molecular mechanisms by which leptin initiates these processes. Here we show for the first time that leptin induces spine formation by activating a TrpC current mediated by the TrpC1 and TrpC3 subunit-containing channels by stimulating translocation of the TrpC channel to the plasma membrane. Furthermore, we show that leptin induction of the TrpC current and subsequent formation of spines requires activation of the CaMKK/CaMK1γ cascade and its downstream targets, β-Pix and Rac 1. A proposed model of this signaling pathway is shown in Figure 8.
Schematic representation of signaling cascade activated by leptin leading to excitatory synapse formation in hippocampal neurons. Jak2, Janus Kinase2; CaMKI, Ca2+/calmodulin kinase I; Rac1, Ras-related C3 botulinum toxin substrate1; Pak1, p21 protein (Cdc42/Rac)-activated kinase 1. The question mark (?) denotes that the source of calcium has not yet been established.
TrpC channels are critical for leptin's actions
Here we demonstrate for the first time that leptin induced spine formation requires activation of a TrpC current in hippocampal neurons. Interestingly, we found that this slow activating current is mediated by TrpC1 and TrpC3 (but not TrpC5) channels and that these subunits are also required for spine formation. TrpC channels also play a critical role in BDNF-dependent spine formation (Amaral and Pozzo-Miller, 2007b; Davare et al., 2009; Li et al., 2012) and TrpC3 (but not TrpC5) channels are similarly required for BDNF induction of both a slow-activating current and spine formation in hippocampal neurons (Amaral and Pozzo-Miller, 2007b), suggesting that BDNF and leptin may use a common pathway. TrkB-IgG, a BDNF scavenger, however did not reduce leptin-induced spine formation (M. Dhar and G. Wayman, unpublished data), suggesting that although the pathways may converge downstream of the respective receptors, leptin's effects are not mediated via release of BDNF.
TrpC3 and TrpC5 channels have been shown to be trafficked to the membrane by BDNF and nerve growth factor (NGF; Bezzerides et al., 2004; Amaral and Pozzo-Miller, 2007b). However, this is the first demonstration of TrpC1 being rapidly trafficked in neurons following treatment with a neurotrophic factor. The TrpC1 channel is primarily proposed to mediate calcium-activated calcium release (Liu et al., 2003; Ambudkar, 2007), although it has been shown to be trafficked to the membrane in response to calcium entry via Orai1 (Cheng et al., 2011). This is also the first demonstration of a role for this channel in spine formation. Interestingly, leptin activates a TrpC4 or 5 channel in the hypothalamus (Qiu et al., 2010; Williams et al., 2011); as our data demonstrates that these subunits are not critical for the leptin-induced TrpC current in hippocampal neurons, leptin appears to activate diverse TrpC currents in different brain regions. One consequence of activating TrpC channels composed of different subunits is that unique kinetic, conductance, and regulatory properties are conferred (Clapham et al., 2005) that could be important for the varied actions of leptin. Although the I–V relationship of the leptin-induced TrpC current we report here is similar to the TrpC current activated by leptin in the hypothalamus (Qiu et al., 2010), it is possible that they have different regulatory or kinetic properties that have not yet been explored.
Leptin uses a novel CaMKK/CaMKI signaling cascade to traffic TrpC channels and induce spine formation
The long isoform of the LepR is a cytokine receptor that can initiate multiple signaling cascades, such as the MAPK, PI3K, and PLC pathway (Myers, 2004). Here we show for the first time that leptin treatment stimulates phosphorylation of CaMKI, which is required for activation of the leptin current as well as leptin-induced spine formation. We also show that it is specifically CaMKKβ and CaMKIγ that are required. Both CaMKI and CaMKK require Ca2+/CaM binding in addition to phosphorylation for activation (Wayman et al., 2008). Although leptin was previously shown to increase intracellular calcium in CA1 neurons (Oomura et al., 2006), the precise source of calcium is as yet unknown. BDNF-induced translocation of TrpC3 in hippocampal neurons requires PLC and release of intracellular calcium stores via IP3 receptors (Amaral and Pozzo-Miller, 2007b). Qiu et al. (2010) also showed that leptin induction of a TrpC current in the hypothalamus requires PLC, although IP3 receptors do not appear to be involved. We also found a requirement for PLC; however, calcium could also enter through other calcium channels, including other TrpC channels or NMDA receptors, as leptin increases calcium entry through NMDA receptors (Shanley et al., 2001) and NMDA receptor activity is vital for leptin's effects on filopodia formation (O'Malley et al., 2007) as is the MEK/Erk pathway. How these pathways interact with activation of the CaMKK/CaMKI cascade is not yet clear. Interestingly, leptin also stimulates phosphorylation of CaMKII (Oomura et al., 2006), which the authors suggest is important for its effects to increase LTP.
Leptin activation of the CaMKK/CaMKI cascade leads to phosphorylation and activation of β-Pix, a Rac1 GEF, and consequently Rac1 activation. We had previously reported that CaMKK, CaMKI, β-Pix, and Rac1 exist in a complex that can respond to synaptic activity to promote the rearrangement of the actin cytoskeleton and therefore stimulate synapse formation (Saneyoshi et al., 2008). CaMKI phosphorylates β-Pix at Ser516 to activate its GEF activity. Here we show that phosphorylation at this site is also critical for the leptin-induced current, trafficking of the TrpC channel and filopodia formation. Given that CaMKIγ also colocalizes with TrpC5 channels in axonal growth cones (Davare et al., 2009), it will be intriguing to determine whether leptin uses the same signaling pathway to activate TrpC4/5 channel-dependent currents in the hypothalamus. CaMKK is also an important regulator of AMP-dependent kinase (AMPK), another critical mediator of leptin's actions (Frühbeck, 2006). Thus, this newly identified signaling cascade also provides a potential mechanism for leptin activation of AMPK.
Finally, we show that Rac1, a Rho family GTPase that is an essential modulator of actin dynamics and synapse formation and maintenance (Nakayama et al., 2000; Ridley, 2001; Penzes et al., 2003), is required for the leptin-induced current, TrpC1 channel trafficking, and filopodia formation. This proposed role of Rac1 is consistent with its established roles in vesicular trafficking (Ridley, 2001; Symons and Rusk, 2003; Chiu et al., 2011) and NGF-induced membrane trafficking of TrpC5 in primary hippocampal cultures (Bezzerides et al., 2004).
Leptin was previously shown to modulate trafficking of other membrane proteins, including both AMPA and NMDA glutamate receptors in hippocampal neurons (Beccano-Kelly and Harvey, 2012). Insertion of postsynaptic glutamate receptors is an essential part for the development of functional synapses on the newly formed dendritic spines and we recently demonstrated that these new spines do form functional synaptic connections when we examine leptin's longer-term effects (Dhar et al., 2014). Leptin may use multiple complementary signaling cascades to promote trafficking of different receptors and channels, as it promotes GluR1 receptor insertion through the inhibition of PTEN, and subsequent increase in PI3K activity (Moult et al., 2010).
Physiological role of leptin actions in the hippocampus
Here we elucidate a novel mechanism by which leptin induces spines in hippocampal neurons, which is thought to be the first step in the formation of new synapses. This role of leptin has consequences beyond development, as mice with deficient leptin signaling (db/db mice) have fewer hippocampal dendritic spines in adulthood in vivo (Stranahan et al., 2009). There is a rich literature showing that hippocampal dendritic spine number correlates with cognitive performance and that changes in spine number are associated with neurological disorders, with too few spines associated with learning and memory defects and depression, and too many spines with schizophrenia and autism (Irwin et al., 2000; Pittenger and Duman, 2008; Hajszan et al., 2009; Kasai et al., 2010). Leptin receptors are found in the CA1, CA3, and DG regions of the hippocampus (Mercer et al., 1996; Shanley et al., 2002; Guo et al., 2012) and leptin has been shown to effect hippocampal neuronal function, synaptic plasticity, and related behaviors (Harvey et al., 2005; Harvey, 2007). Consistent with both a key role of leptin in spine formation and with spine formation being critical for hippocampal-dependent behaviors, mice with deficient leptin signaling (db/db and ob/ob) have deficits in learning and memory tasks and show anxiety and depressive symptoms (Li et al., 2002; Sharma et al., 2010; Guo et al., 2012). By altering the initiation of dendritic spines through the mechanism we demonstrate here, leptin may modulate hippocampal function and so impact hippocampal-dependent behaviors.
In conclusion, our data elucidates a novel molecular pathway by which leptin can activate a TrpC current and initiate spine formation (Fig. 8) and suggests a mechanism by which leptin can initiate synaptic development and modulate hippocampal function fairly rapidly. Leptin has also been shown to increase excitatory synapses onto POMC neurons in the hypothalamus in vivo (Pinto et al., 2004). Leptin activates a TrpC channels in these neurons (Qiu et al., 2010), but it remains to be established whether a similar CaMKK pathway is also involved in leptin-induced synapse formation in other brain regions.
Footnotes
This work was supported by NIH Grant MH086032 (G.A.W.) and DK083452 (S.M.A.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestion and Kidney disease, National Institute of Mental Health, or the NIH.
The authors declare no competing financial interests.
- Correspondence should be addressed to either Dr Gary Wayman or Dr Suzanne Appleyard, Department of IPN, Program in Neuroscience, Washington State University, 100 Dairy Road, Pullman, WA 99164. waymang{at}vetmed.wsu.edu or appleyas{at}vetmed.wsu.edu