Insulin and adipokine signaling and their cross-regulation in postmortem human brain

Abstract

Aberrant insulin and adipokine signaling has been implicated in cognitive decline associated with both Type 2 Diabetes Mellitus (T2D) and neurodegenerative diseases. We established methods that reliably measure insulin, adiponectin and leptin signaling, and their crosstalk, in thawed postmortem mid-frontal cortical tissue from cognitively normal older subjects with a short postmortem interval. Insulin-evoked insulin receptor (IR) activation increases activated, tyrosine-phosphorylated (pY) IRβ on tyrosine residues 960, 1150 and 1151, IRS-1 recruitment to IRβ and phosphorylated AKT1. Adiponectin augments, but leptin inhibits insulin signaling. Adiponectin activates adiponectin receptors to induce APPL1 binding to adiponectin receptor1 (AdipoR1) and 2 (AdipoR2) and T-cadherin and downstream AMPK phosphorylation. Insulin inhibited adiponectin-induced signaling. In addition, leptin-induced leptin receptor (OB-R) signaling promotes JAK2 recruitment to OB-R and JAK2 and downstream STAT3 phosphorylation. Insulin enhanced leptin signaling. These data demonstrate insulin and adipokine signaling interactions in human brain. Future studies can use these methods to examine insulin, adiponectin and leptin metabolic dysregulation in aging and disease states, such as Type 2 Diabetes and Alzheimer’s Disease-Related Dementias (ADRD).

1. Introduction

Type 2 Diabetes Mellitus (T2D) is an established risk factor for cognitive decline and dementia in later life (Cheng et al., 2012; Gudala et al., 2013). The neurobiological mechanisms for this increased risk remain uncertain. Systemic insulin resistance with resultant hyperglycemia best characterizes T2D, most often occurring in the setting of overweight and obesity states. Obesity is associated with derangement in adipocyte release and action of adipokines, including adiponectin and leptin. In peripheral tissues, there is extensive crosstalk between insulin, adiponectin and leptin signaling pathways (Kadowaki et al., 2006; Knights et al., 2014; Kwon and Pessin, 2013; Yadav et al., 2013). However, relatively little is known about the actions of these hormones in the brain (Bloemer et al., 2018; Jovanovic and Yeo, 2010; Thundyil et al., 2012). The contributions from changes and interactions between insulin and adipokine signaling to aging, T2D and various dementias including Alzheimer’s disease (AD) remain largely unknown.

Expression of the receptors for adiponectin and leptin have been shown in the central nervous system (Couce et al., 1997; Kubota et al., 2007; Takeuchi et al., 2000), and adiponectin and leptin are measurable in cerebrospinal fluid (CSF), indicating that these hormones are released from adipose tissue and may communicate directly within the brain (Ebinuma et al., 2007; Nam et al., 2001; Neumeier et al., 2007; Schwartz et al., 1996). Adiponectin and leptin are important regulators of insulin’s actions (Clegg et al., 2006; Kadowaki et al., 2006; Kwon and Pessin, 2013; Paz-Filho et al., 2012). In neural cells and animal models, adiponectin potentiates insulin’s effects through binding to its cognate receptors AdipoR1, AdipoR2 and T-cadherin, leading to recruitment of the signal transducer adaptor protein containing pleckstrin homology domain phosphotyrosine-binding domain and leucine zipper motif (APPL1) (Cheng et al., 2012; Deepa and Dong, 2009; Mao et al., 2006; Ruan and Dong, 2016). These initial signaling events activate adenosine monophosphate dependent protein kinase (AMPK), peroxisome proliferator-activated receptor-α (PPAR-α), p38 mitogen-activated protein kinase (p38MAPK) and probably other as-of-yet unknown signaling pathways (Hug et al., 2004; Thundyil et al., 2012; Yamauchi et al., 2007). Adiponectin has been shown to reduce inflammation and oxidative stress, thereby imparting cell protection (Esmaili et al., 2014; Park et al., 2008; Shrestha and Park, 2016). Reduction of adiponectin in obesity is associated with insulin resistance, dyslipidemia, and atherosclerosis in humans (Nigro et al., 2014). Nevertheless, the alterations to adiponectin and its receptors in human CNS during TD2 or the pathogenesis of brain disorders and their contributions to brain insulin resistance, as in AD, remain elusive (Arnold et al., 2018; Talbot et al., 2012; Yarchoan and Arnold, 2014).

While adiponectin secretion is reduced in obese subjects, circulating leptin levels, in contrast, are increased (Considine et al., 1996; Maffei et al., 1995). Leptin suppresses appetite, promotes thermogenesis, enhances fatty acid oxidation, decreases glucose levels and reduces body weight and fat (Amitani et al., 2013; Paz-Filho et al., 2012). Although some epidemiological data suggest a negative influence of leptin on brain function (Whitmer et al., 2005), neuroprotective effects of leptin have also been observed in animals (Signore et al., 2008; Tezapsidis et al., 2009; Wan et al., 2015). There is evidence that glucose uptake by different tissues following leptin treatment is altered in a tissue-specific manner (Minokoshi et al., 1999). Given that most studies on the crosstalk between insulin and adipokines have been conducted in peripheral tissues of rodents (Berthou et al., 2011; Cusin et al., 1998; Pelleymounter et al., 1995; Pérez et al., 2004), it is essential to assess the interactions between insulin and adipokines directly in human brain tissues, especially in diseases that are poorly modeled in animals, such as AD.

In this study, we describe an ex vivo stimulation method to study insulin, adiponectin and leptin signaling, and their interactions, in human postmortem brain tissues. This method has been modified and advanced from an earlier method previously described by our group that was used successfully to study the signaling of various receptors (Hahn et al., 2006; Talbot et al., 2012; Wang et al., 2012; Wang et al., 2009; Wang et al., 2017). This method directly and reliably detected insulin, adiponectin and leptin signaling, as well as their interactions. Implementing this method in disease settings such as Alzheimer’s disease-Related Dementias (ADRD) will help identify disease-relevant mechanisms as potential targets for effective interventions.

2. Methods

2.1. Study approval.

The human brain study protocol using postmortem human brain tissues conformed to the tenets of the Declaration of Helsinki as reflected in a previous approval by the City College of New York and City University of New York Medical School human research committee. It has been determined that this project does not meet the definition of human subject research as defined by the federal regulations (45 CFR 46.102(d) (f)) and therefore no further IRB review or approval is required.

The animal procedures comply with the National Institutes of Health Guide for Care Use of Laboratory Animals and were approved by the City College of New York Animal Care and Use Committee (IACUC) - Protocol#:987.1.

2.2. Description of subjects and postmortem intervals.

This study was derived from the “Brain Insulin Resistance in Aging” (BIRA) study, funded by the National Institutes of Health (NIH; see Acknowledgements). Descriptive information on the brain specimens, subjects, and associated data studied are provided in the Table 1.

Table 1.

Description of subjects and postmortem intervals

	BIRA subset for ex vivo stimulation
	Without MCI or dementia and diabetes	With MCI or dementia or diabetes
Number of cases (n)	15	63
Experimental methods study	included	excluded
History of diabetes (n, %)	0 (0.00%)	39 (61.90%)
Age at death, years (mean, STD)	87.02 (6.21)	87.23 (6.01)
Postmortem intervals, hours (mean, STD)	5.26 (2.46)	4.74 (2.05)
Sex, women (n, %)	7 (42.86%)	27 (42.86%)
White non-Latino (n, %)	15 (100.00%)	57 (90.48%)
Education, years (mean, STD)	19.07 (3.37)	18.16 (3.36)

The 15 cases included in this study are part of the subset of 78 cases from the Brain Insulin Resistance in Aging (BIRA) study in which ex vivo stimulation data are available. In the total sample used in the BIRA study (n = 150), cases were autopsied subjects (75 with diabetes and 75 without), randomly selected from large ongoing longitudinal clinical-pathologic study of aging and dementia, the Religious Orders Study (ROS). The research volunteers from ROS are community-dwelling, older (nuns, priests, and brothers) from across the United States, without known dementia at study entry, who agree to annual clinical evaluations, including detailed neuropsychological testing, and brain donation at time of death. The follow-up rate in ROS is 95%, and the autopsy rate exceeds 90% (Bennett et al., 2018). Of the 150 BIRA subjects, a total of 78 subjects had PMI <12 hours and, among these, 15 subjects for this study were specifically selected among those with normal cognition and without diabetes, and without significant neurodegenerative or other neuropathology (De Jager et al., 2018). The 15 subjects in this experimental study did not differ with the others (n = 63) on age at death (t(76)=0.12, p =0.902),PMI (t(76)= −0.85, p = 0.400), education (t(76)= −0.94, p =0.350), sex and race (both fisher exact, p>0.99).

2.3. Validation tests of the ex vivo method in human postmortem tissue.

Dose-response effects of insulin, adiponectin, and leptin were tested in approximately 10 mg of postmortem human dorsolateral prefrontal cortical (DLPFC) slices (100 μm × 100 μm × 3 mm) derived from 3 cognitively normal control subjects (2 females and 1 male) with mean age at death= 84.8 (± 3.4) years and postmortem intervals (PMIs) of 7.4 (± 1.2) hours. To assess the dose-dependency of the insulin effects, DLPFC slices were incubated with 0, 0.1, 1, and 10 nM insulin in oxygenated Kreb’s-Ringer at 37°C for 30 min as described previously in our earlier studies (Talbot et al., 2012; Wang et al., 2012). The insulin signaling was determined by the levels of IRS-1 recruited to IRβ and phosphorylation of IRβ and AKT1. Similar conditions were used to test the adiponectin and leptin signaling by incubating the slices with 0, 1, 10, and 100 μg/ml recombinant human adiponectin and with 0, 1, 10, and 100 nM recombinant human leptin, respectively. The adiponectin signaling was assessed by the levels of APPL1-coupled AdipoR1, AdipoR2 and T-cadherin, as well as phosphorylation of AMPKα1/2. The leptin signaling was measured by the levels of Janus kinase 2(JAK2)-associated leptin receptors, OBRs and phosphorylation of JAK2 and signal transducer and activator of transcription 3 (STAT3).

PMI effects on insulin signaling and its regulation by adipokines, adiponectin and leptin were measured by the responses to 1 nM insulin alone and in combination with 10 μg/ml adiponectin or 10 nM leptin in the postmortem prefrontal cortex of 2.5 month-old male Sprague-Dawley rats (Taconic Farms) and conducted similarly to the test described previously (Talbot et al., 2012). The prefrontal cortex is defined as the medial frontal pole cortex (pregenual) that contains limbic and infralimbic regions. In compliance with the NIH Guide for the Care and Use of Laboratory Animals, the animals were sacrificed by quick CO₂ asphyxiation. Brains were removed quickly on ice at death (PMI, 0 hours) or remainded in the intact body at 4°C for 4, 8, 12 or 16 hours, simulating human postmortem conditions. In other animals, the intact body was kept at 25°C for 4 hours before brain extraction. Upon removal, brains were frozen and stored at −80°C for 4 weeks until the experimental days.

2.4. Brain slice preparation.

Human fresh-frozen tissue from the DLPFC was gradually thawed from −80°C to −20°C by placing the tissues in −20°C freezer for approximately 10 minutes, remove about 20 mg section on ice bed and cut into 100 μm × 100 μm × 3 mm slices using a chilled McIlwain tissue chopper. Slices equaling about 20 mg of tissue were suspended in 1 ml ice-cold oxygenated Krebs-Ringer solution (K-R) containing 25 mM HEPES (pH 7.4), 118 mM NaCl, 4.8 mM KCl, 1.3 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 25 mM NaHCO₃, 10 mM glucose, 100 μM ascorbate, EDTA-free protease inhibitor cocktail (Roche, 04693159001) and 0.01 U/ml soybean trypsin inhibitor and centrifuged briefly. After two additional washes with 1 ml ice-cold K-R, brain slices were suspended in 1 ml K-R.

2.5. Assessment of Ca²⁺ influx as the indices of mitochondrial function and cell death.

Since the level of voltage-gated Ca²⁺ channel activity indicates the physiological integrity of the cells, depolarization-induced Ca²⁺ influx in the postmortem rat prefrontal cortex slices was used to assess the PMI effects on cell integrity. The PMI effects on depolarization-induced Ca²⁺ influx and insulin signaling were then compared to ascertain whether postmortem tissues retain sufficient mitochondrial function and responsiveness to insulin stimulation. As described previously with slight modification (Wang et al., 2012; Wang et al., 2010), approximately 5 mg of thawed postmortem rat prefrontal cortical slices (100-μm × 100-μm × 3-mm) were incubated at 37°C for 5min in oxygenated K-R that have 1.3 mM Ca²⁺ containing 5 μCi ⁴⁵Ca²⁺ (10 Ci/mmol) followed by incubation with vehicle, or 65 mM K⁺ (made with isomolar replacement of Na⁺) for 1 min in a total incubation volume of 0.5 ml. The reaction was terminated by the addition of 0.5 ml of ice-cold Ca²⁺-free K-R containing 0.5 mM EGTA and centrifugation at 4°C. After two additional washes, ⁴⁵Ca²⁺ contents in brain slices were solubilized in 1% Triton-X100 and assessed using scintillation spectrometry (Beckman). Background ⁴⁵Ca²⁺ was estimated using lysed brain slices. Basal Ca²⁺ influx was defined as the ⁴⁵Ca²⁺ levels in the Kreb’s-Ringer incubated brain slices subtracting background ⁴⁵Ca²⁺ count. Depolarization-induced Ca²⁺influx was defined as the ⁴⁵Ca²⁺ contents in the 65 mM K⁺-incubated brain slices subtracting background ⁴⁵Ca²⁺ count. The percentage increase in Ca²⁺ influx was calculated as percentage [(K⁺-depolarization treated-Kreb’s-Ringer)/Kreb’s-Ringer].

2.6. Ex vivo insulin stimulation.

Approximately 20 mg of thawed postmortem human DLPFC slices were incubated for 30 minutes at 37°C with vehicle (K-R) alone or with vehicle plus recombinant human insulin (Invitrogen 12585–014), recombinant human adiponectin (#4901, Biovision) or recombinant human leptin (CYT-338, Prospec). The incubation mixture (total volume, 0.5 ml) was aerated every 10 minutes with 95% O₂ and 5% CO₂ for 1 minute. Signaling was terminated with 1.5 ml ice-cold Ca²⁺-free K-R. The tissue was collected by brief centrifugation and homogenized in 250 μl ice-cold immunoprecipitation buffer (25 mM HEPES, pH 7.5; 200 mM NaCl; 1 mM EDTA; protease inhibitor cocktail (Roche), 5 mM NaF; 1 mM sodium vanadate; 0.5 mM β-glycerophosphate; and 0.02% 2-mercaptoethanol). Homogenates were centrifuged at 1,000 g for 5 minutes at 4°C, and the resultant supernatant (post-mitochondrial fraction) was sonicated for 10 seconds on ice. Proteins were then solubilized in immunoprecipitation buffer containing 0.5% digitonin, 0.2% sodium cholate, and 0.5% NP-40 for 60 minutes at 4°C with end-to-end rotation. The resultant lysates were then cleared by centrifugation at 50,000 g for 5 minutes and diluted with 750 μl immunoprecipitation buffer. Protein concentrations were measured using the Bradford method (Bio-Rad).

2.7. Immunoprecipitation.

Insulin, adiponectin, and leptin signaling were each assessed separately by isolating the proteins of interest, 200 μg of tissue lysates were immunoprecipitated overnight at 4°C with 1.0–1.5 μg of antibodies to IRβ and AKT1 (insulin signaling), APPL1and AMPKα1/2 (adiponectin signaling), and JAK2 and STAT3 (leptin signaling) covalently immobilized onto protein A/G–agarose beads (Pierce Thermo). The antibodies used are shown in Table 2A. Immunoprecipitates were incubated with 75 μl antigen elution buffer pH2.8 (Pierce Thermo) and 2% SDS for 2 minutes at room temperature, centrifuged to remove antibody–protein A/G–agarose complexes, and neutralized immediately with 10 μl of 1.5 M Tris buffer (pH 8.8) followed by addition of 65 μl 2× PAGE sample buffer and boiled for 5 minutes. Supernatants were probed for any residual target proteins, which showed that greater than 90% of the target proteins were immunoprecipitated in all subjects (Figure 1).

Table 2.

Antibodies used for immunoprecipitation

Antigen	Antibody (Ab)	Ab Type	Ab/200 μg lysate
AKT1	Santa Cruz 5298	Mouse mAb	1.2 μg
AMPKα1/2	Santa Cruz 74461	Mouse mAb	1.0 μg
APPL1	Santa Cruz 67402	Rabbit pAb	1.2 μg
JAK2	Santa Cruz 294	Rabbit pAb	1.2 μg
IRβ	Santa Cruz 20739	Rabbit pAb	1.0 μg
STAT3	Santa Cruz 7179	Rabbit pAb	1.0 μg

Key: AMPK, adenosine monophosphate-dependent protein kinase; APPL1, adaptor protein containing pleckstrin homology domain phosphotyrosine-binding domain and leucine zipper motif; JAK2, Janus kinase 2; IR, insulin receptor; STAT3, signal transducer and activator of transcription 3.

Figure 1.

Representative Western blots verifying completeness and specificity of immunoprecipitation for the frontal cortical proteins studied in postmortem human brains: IRβ (A), AKT1 (B), APPL1 (C), AMPKα1/2 (D), JAK2 (E) and STAT3 (F). Three postmortem human frontal cortices from control subjects were chosen for testing. Results shown are the representative of three samples. The lanes in each blot show relative amount of antigen in the tissue lysate, the immunoprecipitate (IP) of that lysate, and the remaining supernatant. The solubilized lysate was used to assess the specificity of immunoprecipitation using indicated antibody, IgG and protein A/G beads. Further, the resultant supernatant following IP was IP with specific antibodies. Results show at least 90% of each antigen was immunoprecipitated and each antibody was specific to the antigen indicated. N = 3

2.8. Western blotting.

Solubilized immunoprecipitates derived from 100 μg brain slice lysates were size fractionated on 7.5% or 10% SDS-PAGE, and then electrophoretically transferred to nitrocellulose membranes. Membranes were washed with PBS, blocked for 2 hours at room temperature with 10% non-fat milk in PBS containing 0.1% Tween-20 (PBST), and washed 3× in 0.1% PBST (2 minutes each). To assess protein activation, membranes loaded with the above-described immunoprecipitates were incubated overnight at 4°C with antibodies (Table 3) to pY^1150/1151IRβ, pY⁹⁶⁰IRβ, IRS-1, pS⁴⁷³AKT1, pT³⁰⁸AKT1, AdipoR1, AdipoR2, T-cadherin, pT^183/172AMPKα1/2, APPL1, pY^1007/1008JAK2, Ob-R, or pY⁷⁰⁵STAT3. To assess total protein levels as well as IRS-1 recruited to IRβ, membranes were stripped, washed and blocked with 10% non-fat milk and re-probed overnight at 4°C with antibodies (Table 2B) to IRβ, IRS-1, AKT1, AdipoR2, APPL1, JAK2, or STAT3. After initial probing and again after re-probing, membranes were washed 3× in 0.1% PBST (2 minutes each), incubated for 1 hour with 1:5,000 dilution of species-appropriate, HRP-conjugated secondary antibodies, and washed 3× in 0.1% PBST baths (2 minutes each). Immunoreactivity was visualized by reaction in ECL Plus chemiluminescent reagent for exactly 5 minutes, followed by exposure of the blot to X-ray film. Bands at the relevant molecular masses were quantified using a GS-800 calibrated densitometer (Bio-Rad Laboratories). Ratios of phosphorylated to total levels of each antigen were calculated.

Table 3.

Antibodies used for western blotting

Antigen	Antibody (Ab)	Ab Type	Ab dilution
AdipoR1	Santa Cruz 518,030	Mouse mAb	1:1000
AdipoR2	Santa Cruz 514045	Mouse mAb	1:1000
AKT1–3	Santa Cruz 8312	Rabbit pAb	1:500
pS473AKT1–3	Santa Cruz 514032	Mouse mAb	1:750
pT308AKT1–3	Santa Cruz 271964	Mouse mAb	1:750
AMPKα1/2	Santa Cruz 74461	Mouse mAb	1:750
pT183/172AMPKα1/2	Abcam 23875	Rabbit pAb	1:1000
APPL1	Santa Cruz 271,909	Mouse mAb	1:750
IRβ	Santa Cruz 81465	Mouse mAb	1:500
pY960 IRβ	Invitrogen 44–800G	Rabbit pAb	1:1000
pY1150/1151IRβ	Santa Cruz 81500	Mouse mAb	1:1000
IRS-1	Santa Cruz 8038	Mouse mAb	1:750
JAK2	Santa Cruz 390,539	Mouse mAb	1:750
pY1007/1008JAK2	Cell Signaling #3776	Rabbit mAb	1:1000
OB-R	Santa Cruz 8391	Mouse mAb	1:750
STAT3	Santa Cruz 8019	Mouse mAb	1:750
pY705STAT3	Santa Cruz 8059	Mouse mAb	1:750
T-Cadherin	Santa Cruz 166875	Mouse mAb	1:750

Key: AdipoR1, adiponectin receptor 1; AdipoR2, adiponectin receptor 2; AMPK, adenosine monophosphate-dependent protein kinase; APPL1, adaptor protein containing pleckstrin homology domain phosphotyrosine-binding domain and leucine zipper motif; OB-R, leptin receptor; JAK2, Janus kinase 2; IR, insulin receptor; STAT3, signal transducer and activator of transcription 3; pY, phosphorylated; IRS-1, insulin receptor substrate-1.

3. Results

3.1. The completeness and specificity of immunoprecipitation.

The completeness and specificity of immunoprecipitation for each antigen protein of interest in the insulin, adiponectin and leptin signaling cascades were assessed. In Figure 1, each antigen was demonstrated in representative Western blots to verify that immunoprecipitation with anti-IRβ (A), -AKT1 (B), -APPL1 (C), -AMPKα1/2 (D), -JAK2 (E) and -STAT3 (F) was complete and specific. The completeness of immunoprecipitation was shown by the comparable amount of antigen in the tissue lysate and the immunoprecipitate (IP) of that lysate with negligible antigen observed in the remaining supernatant (Left panel in each group). As the supernatants of the IP show, greater than 90% of each antigen was immunoprecipitated. The specificities of the antibodies for IRβ (A), -AKT1 (B), -APPL1 (C), -AMPKα1/2 (D), -JAK2 (E) and -STAT3 (F), were demonstrated by the data showing that non-immune rabbit IgG and protein A/G beads brought down negligible amounts of IRβ, AKT1, APPL1, AMPKα1/2, JAK2 and STAT3 (Right upper panel in each group). The fact that the remaining supernatant from non-immune rabbit IgG and protein A/G beads recovered comparable amounts of IRβ and AKT1 with the anti-IRβ and anti-AkT1 immunoprecipitates further confirms the completeness of these immunoprecipitations (Right lower panel in each group).

3.2. Insulin, adiponectin, and leptin dose-dependently activate their cognate receptors in human postmortem brain tissues.

To test the effects of insulin in activating insulin receptor, thawed postmortem DLPFC slices from three cognitively normal elderly subjects were incubated ex vivo with 0.1 – 10 nM of human recombinant insulin for 30 min. Lysate prepared from the Kreb’s-Ringer and insulin incubated DLPFC slices showed that 0.1 – 10 nM insulin induced clear concentration-dependent activation of insulin receptor signaling. Increasing insulin concentrations resulted in increasing phosphorylation of the tyrosine residues: Y¹¹⁵⁰, Y¹¹⁵¹ and Y⁹⁶⁰ in the IRβ leading to recruitment of IRS-1 (Figure 2) and activation of the downstream signaling molecule, AKT1, indicated by an increase in phosphorylation of the serine (pS⁴⁷³) and threonine (pT³⁰⁸) residues in AKT1. These results suggest that stimulation with 1 nM of insulin is adequate for evaluating IR responses in the follow-up studies (Figure 2).

Figure 2.

Dose-dependency of insulin signaling in postmortem human frontal cortices.

Frontal cortical slices from neurologically normal subjects with postmortem interval < 12 hours were used to determine insulin signaling induced by 0.1 – 10 nM recombinant human insulin. Following incubation with varying concentrations of insulin for 30 min at 37°C, brain slices were collected, solubilized and immunoprecipitated with immobilized anti-IRβ or -AKT1 antibodies. The contents in resultant immunoprecipitates were analyzed by Western blotting. Insulin dose-dependently increased phosphorylation of tyrosine residues: 1150/1151 and 960 of the insulin receptor β subunit (IRβ): pY^1150/1151 and pY⁹⁶⁰ as well as recruitment of insulin receptor substrate-1 (IRS-1) to IRβ (A) and phosphorylation of AKT1 at the serine⁴⁷³ (by mTOR2) and threonine³⁰⁸ (by PDK1) (B). The blots were stripped and re-probed with anti-IRβ or -AKT1 to ascertain comparable IRβ and AKT1 were immunoprecipitated, respectively. These data indicate that postmortem brain tissues with short postmortem interval are responsive to insulin can be reliably used to assess changes in insulin receptor activities and signaling. N = 3

The effects of adiponectin on activating adiponectin receptor signaling was tested in DLPFC slices by ex vivo incubation with 1–100 μg/ml human recombinant adiponectin. In response to varying concentrations of adiponectin, association of APPL1 with AdipoR1, AdipoR2 and T-cadherin was increased in a concentration-dependent manner. The adiponectin dose-response curve for the increased APPL1 and adiponectin receptor complexes parallels that of activation of the downstream AMPKα1/2 indicated by increasing pT^183/172AMPKα1/2 (Figure 3). These data collectively indicate 10 μg/ml of adiponectin, a concentration well within physiological levels, is suitable for the assessment of adiponectin-elicited effects in the later investigations.

Figure 3.

Dose-dependency of adiponectin signaling in postmortem human frontal cortices.

Frontal cortical slices from neurologically normal subjects with postmortem interval < 12 hours were used to determine insulin signaling induced by 1 – 100 μg/ml recombinant human adiponectin. Following incubation with varying concentrations of adiponectin for 30 min at 37°C, brain slices were collected, solubilized and immunoprecipitated with immobilized anti-APPL1 or -AMPKα1/2 antibodies. The contents in resultant immunoprecipitates were analyzed by Western blotting. Adiponectin dose-dependently increased association of the adiponectin receptor subtypes: AdipoR1, AdipoR2 and T-cadherin with signaling adaptor protein, APPL1 (A) as well as in the phosphorylation of AMPKα1/2 at the threonine¹⁸³ and threonine¹⁷² of the AMPK (B). These data indicate that postmortem brain tissues with short postmortem interval are responsive to insulin can be reliably used to assess changes in adiponectin receptor signaling. N = 3

The effects of leptin on leptin receptor (OB-R) signaling was assessed in DLPFC slices by ex vivo incubation with 1–100 nM recombinant human leptin. Exposure to 1–100 nM of leptin led to dose-dependent increases in coupling of long- and short-form OB-Rs with JAK2 and activation of JAK2 and STAT3 as indicated by the increases in pY^1007/1008JAK2 and pY⁵⁰⁵STAT3, respectively (Figure 4). These data indicate that 10 nM of leptin is appropriate can be used to probe the biological effects of leptin.

Figure 4.

Dose-dependency of leptin signaling in postmortem human frontal cortices.

Frontal cortical slices from neurologically normal subjects with postmortem interval < 12 hours were used to determine insulin signaling induced by 1 – 100 nM recombinant human leptin. Following incubation with varying concentrations of leptin for 30 min at 37°C, brain slices were collected, solubilized and immunoprecipitated with immobilized anti-JAK2 or -STAT3 antibodies. The contents in resultant immunoprecipitates were analyzed by Western blotting. Adiponectin dose-dependently increased association of OB-Rs with and activation (tyrosine phosphorylation) of JAK2, a non-receptor tyrosine kinase (A) as well as activation (tyrosine phosphorylation) of STAT3 (B). These data indicate that postmortem brain tissues with short postmortem interval are responsive to insulin can be reliably used to assess changes in leptin receptor signaling. N = 3

3.3. Insulin signaling and its regulation by adipokines are demonstrated in brain tissue with short PMI in a simulating postmortem rat model.

Ex vivo tests of the PMI effects on insulin signaling and its regulation by the adipokines were conducted in four individual sets of postmortem rat prefrontal cortex slices with PMIs ranging from 0 (snap-frozen) to 16 hours. Insulin induced clear and reliable activation of the IR when the PMIs were under 12 hours (Figure 5A). Within 12-hour PMIs, adiponectin potentiated but leptin suppressed this insulin-evoked IR activation. The significantly elevated basal activity and diminished responsiveness to insulin and adipokines led to dramatic reduction in insulin signaling and its regulation by adipokines at the 16-hour time point.

Figure 5.

Effects of postmortem delays on insulin signaling and adipokine regulation as well as basal and K⁺-depolarization evoked calcium influx using postmortem rat frontal cortices.

(A). Sample blots on the rat frontal cortices showing that PMIs up to 12 hours had no significant effects on basal levels of IRβ, IRβ activation as indicated by the pY^1150/1151 levels evoked by 1 nM insulin or effects of adiponectin (10 μg/ml) and leptin (10 nM). An elevated basal IRβ and reduced IRβ activation as well as adiponectin and leptin effects was observed at 16 hours PMI. (B) PMI effects on mitochondrial function and cell viability were tested by assessing basal and 65mM K⁺-evoked Ca influx in 5 μCi ⁴⁵Ca²⁺ (10 Ci/mmol) incubated rat prefrontal cortical slices followed by incubation with Kreb’s Ringer, or 65 mM K⁺ (made with isomolar replacement of Na⁺) for 1 min. The Ca uptake was indicated by the levels of ⁴⁵Ca²⁺ retained in the brain slices assessed by scintillation spectrometry. N = 4 per PMI

*p < 0.01, **p < 0.05 Compared to respective non-stimulated Kreb’s Ringer baseline for each PMI.

#p < 0.01, ##p < 0.05 Compared to basal ⁴⁵Ca²⁺ level insulin stimulation alone.

Since the biological activities of insulin and adipokines depend heavily on intact mitochondria, we tested the effects of different PMIs on cell mitochondrial integrity in the postmortem rat prefrontal cortex slices using depolarization-induced ⁴⁵Ca²⁺ influx that is highly sensitive to mitochondrial changes. In response to 30-second 65mM K⁺-evoked depolarization, ⁴⁵Ca²⁺ influx via the voltage-gated Ca channels was markedly higher than in the non-stimulated Kreb’s-Ringer condition (Figure 5B). However, the non-stimulated ⁴⁵Ca²⁺ levels started to increase in the 8-hour PMI samples and further doubled at 16 hours, indicating reduced cell mitochondrial integrity. Simultaneously, we also observed a reduction in K⁺-evoked depolarization-induced ⁴⁵Ca²⁺ influx when PMI reached 12 hours and, more prominently, at 16 hours. These data therefore suggest that postmortem human brain tissues obtained with PMIs within 12 hours have sufficient cell mitochondrial integrity for the study of insulin, adiponectin, and leptin signaling (Figure 5A and 5B).

3.4. Insulin, adiponectin, and leptin signaling and their regulation in postmortem frontal cortices from cognitively normal human subjects.

Insulin signaling and its regulation by adiponectin and leptin were assessed by ex vivo incubation of human DLPFC slices from 15 autopsied subjects (mean age at death = 87.02 ± 1.60 years; 8 women) without neurodegenerative or other CNS diseases, or diabetes mellitus. We used 1 nM insulin alone or together with 10 μg/ml adiponectin or 10 nM leptin for 30 min at 37°C in oxygenated Kreb’s-Ringer. We then analyzed the lysates derived from these slices for insulin signaling by anti-IRβ and anti-AKT1 immunoprecipitation followed by Western blotting. In accordance with our earlier work (Talbot et al., 2012), 1 nM insulin activated insulin signaling evidenced as a 4.4–4.7 fold increase in pY^1150/1151IRβ, pY⁹⁶⁰IRβ and IRS-1 bound IRβ (Figure 6A). Insulin-induced insulin signaling was also supported by a 4.2-fold increase in activated AKT1 (pS⁴⁷³and pT³⁰⁸) (Figure 6B). Co-incubation of insulin with adiponectin resulted in 20.4 ± 3.4 – 28.2 ± 3.0% potentiation in insulin-evoked increases in pY^1150/1151IRβ, pY⁹⁶⁰IRβ and IRS-1 bound IRβ (Figure 6A) and 25.8 ± 3.1 – 28.1 ± 3.1% potentiation in insulin-induced increases in pS⁴⁷³- and pT³⁰⁸- AKT1 (Figure 6B). In contrast, co-incubation with leptin decreased insulin-evoked increases in pY^1150/1151IRβ, pY⁹⁶⁰IRβ and IRS-1 bound IRβ by 17.6 ± 1.7 – 18.1 ± 2.5% (Figure 6A) and insulin-induced increases in pS⁴⁷³- and pT³⁰⁸- AKT1 by 15.2 ± 2.2 – 17.9 ± 2.7% (Figure 6B).

Figure 6.

Direct demonstration of adiponectin enhances and leptin suppresses insulin receptor activation and signaling in postmortem human brain.

Frontal cortical slices from neurologically normal subjects with postmortem interval < 12 hours were used to assess insulin signaling and its regulation by adiponkines: adiponectin and leptin at the near physiological concentrations. Incubation with insulin alone for 30 min at 37°C leads to phosphorylation of tyrosine residues: 1150/1151 and 960 of the insulin receptor β subunit (IRβ): pY^1150/1151 and pY⁹⁶⁰ as well as recruitment of insulin receptor substrate-1 (IRS-1) to IRβ (A, B) and phosphorylation of AKt1 at the serine⁴⁷³ (by mTOR2) and threonine³⁰⁸ (by PDK1) (C, D). Addition of Adiponectin (10 μg/ml) and leptin (10 nM) together with insulin enhance and suppress insulin signaling, respectively (A, B, C, D). These data indicate that postmortem brain tissues can be reliably used to assess changes in insulin receptor activities and regulation in pathological conditions. N = 15

*p < 0.01 Compared to non-stimulated Kreb’s Ringer baseline.

#p < 0.01 Compared to insulin stimulation alone.

To assess adiponectin signaling and its regulation by insulin and leptin, postmortem human DLPFC slices were incubated with 10 μg/ml exogenous adiponectin alone or together with 1 nM insulin or 10 nM leptin for 30 min at 37°C in oxygenated Kreb’s-Ringer. The lysates derived were analyzed for adiponectin signaling by the levels of APPL1-associated adiponectin receptors AdipoR1, AdiopR2 and T-Cadherin as well as the activated downstream AMPKα1/2 in the anti-APPL1 and anti-AMPKα1/2 immunoprecipitates, respectively, using Western blots. Adiponectin (10 μg/ml) increased APPL1 -AdipoR1, AdiopR2 and T-Cadherin complexes by 4.6–4.7 fold (Figure 7A). Adiponectin at 10 μg/ml also induced a 4.7-fold increase in activated (pT^183/172) AMPKα1/2 (Figure 7B). Co-incubation of insulin but not leptin with adiponectin resulted in 18.6 ± 2.8 – 19.6 ± 3.5% reduction in adiponectin-evoked increases in the levels of the APPL1 - AdipoR1, AdiopR2 and T-Cadherin complexes (Figure 7A) and a 20.2 ± 4.4% decrease in the adiponectin-evoked increases in pT^183/172AMPKα1/2 (Figure 7B).

Figure 7.

Direct demonstration of insulin suppresses adiponectin signaling in postmortem human brain.

Frontal cortical slices from neurologically normal subjects with postmortem interval < 12 hours were used to assess signaling of the adiponectin receptor and its regulation by insulin and leptin at the near physiological concentrations. Incubation with adiponectin (10 μg/ml) alone for 30 min at 37°C promotes adiponectin signaling as illustrated by increases in association of the adiponectin receptor subtypes: AdipoR1, AdipoR2 and T-cadherin with signaling adaptor protein, APPL1 (A, B) as well as in the phosphorylation of AMPKα1/2 at the threonine¹⁸³ and threonine¹⁷² of the AMPK (C, D). Addition of insulin (1 nM) together with adiponectin reduces adiponectin signaling as indicated by lower AdipoR1, AdipoR2 and T-cadherin levels as well as pT^183/172 AMPKα1/2 than adiponectin alone (A, B, C, D). In contrast to insulin, leptin (10 nM) is without an effect on adiponectin signaling. These data indicate phosphorylation of AMPKα1/2 in the postmortem brain tissues can be reliably used to assess changes in adiponectin signaling cascade and regulation in pathological conditions. N = 15

*p < 0.01 Compared to non-stimulated Kreb’s Ringer baseline.

#p < 0.01 Compared to insulin stimulation alone.

To assess the leptin signaling and its regulation by insulin and adiponectin, postmortem human DLPFC slices were incubated with 10 nM exogenous leptin individually or together with 1 nM insulin or 10 μg/ml adiponectin for 30 min at 37°C in oxygenated Kreb’s-Ringer. The lysates derived from these slices were then analyzed for leptin signaling by the levels of activated (pY^1007/1008)JAK2 and JAK2-associated leptin receptors, OB-Rs as well as the activated (pY⁷⁰⁵) downstream STAT3 in the anti-JAK2 and anti-STAT3 immunoprecipitates, respectively, using Western blots. Leptin (10 nM) increased pY^1007/1008JAK2 and JAK2-coupled long- and short-form of OB-R (long and short forms) by 4.2–4.3 folds (Figure 8A). In accord, 10 nM leptin at 10 nM also induced a 4.3-fold increase in pY⁷⁰⁵STAT3 (Figure 8B). Co-incubation of leptin with insulin but not adiponectin increased leptin signaling as indicated by 21.8 ± 2.2%, 24.6 ± 2.9% and 20.5 ± 3.1% increases in leptin-evoked increases in the levels of pY^1007/1008JAK2, JAK2-coupled OB-R long- and short-form, respectively (Figure 8A). In support, insulin also increased leptin-induced pY⁷⁰⁵STAT3 by 22.1 ± 2.5% (Figure 8B).

Figure 8.

Direct demonstration of insulin potentiates leptin signaling in postmortem human brain.

Frontal cortical slices from neurologically normal subjects with postmortem interval < 12 hours were used to assess signaling of the leptin receptor (OB-R) and its regulation by insulin and adiponectin at the near physiological concentrations. Incubation with leptin (10 nM) alone for 30 min at 37°C increases association of OB-Rs with and activation (tyrosine phosphorylation) of JAK2, a non-receptor tyrosine kinase (A, B) as well as activation (tyrosine phosphorylation) of STAT3 (C, D). Addition of insulin (1 nM) but not adiponectin (10 μg/ml) together with leptin increases leptin signaling as indicated by a higher OB-R, pY^1007/1008JAK2 and pY⁷⁰⁵STAT3 levels than leptin alone (A, B, C, D). These data indicate that JAK2 association with OB-Rs and activation of JAK2 in the postmortem brain tissues can be reliably measured and used to assess changes in leptin signaling pathway and regulation in pathological conditions. N = 15

*p < 0.01 Compared to non-stimulated Kreb’s Ringer baseline.

#p < 0.01 Compared to insulin stimulation alone.

4. Discussion

In this study, we experimentally demonstrated insulin, adiponectin and leptin signaling in the postmortem human brain. We further showed that adiponectin enhances, and leptin inhibits, insulin signaling in human brain. This finding is similar to what has been shown in peripheral tissues (Kadowaki et al., 2006; Knights et al., 2014; Kwon and Pessin, 2013; Yadav et al., 2013; Yates et al., 2012). Less established is how insulin conversely regulates adiponectin and leptin signaling pathways, although there is evidence that insulin increases adiponectin and leptin synthesis and secretion from adipose cells (Blümer et al., 2008; Hajri et al., 2010; Marques-Oliveira et al., 2018; Tsubai et al., 2016). Here we show that insulin inhibits adiponectin and enhances leptin signaling, suggesting an auto-regulatory feedback loop within this hormonal brain network. To our knowledge, this has not previously been directly examined in either brain or peripheral tissues in human or other systems.

In accord with our previous work (Talbot et al., 2012; Wang et al., 2012; Wang et al., 2017), ex vivo insulin induced clear, reliable, dose-dependent activation of the insulin signaling pathway. Increasing insulin doses promoted tyrosine phosphorylation (pY) of the IRβ kinase regulatory domain (Y1150, Y1151) and IRS-1 docking site (Y960) and recruitment of IRS-1 to IRβ, as well as activation of downstream signaling molecule, pS⁴⁷³- and pT³⁰⁸- AKT1. To assess the effect of adiponectin and leptin on insulin responses, we selected the 1 nM concentration, close to physiological levels of brain insulin (Baskin et al., 1987; LE ROITH et al., 1983).

Although adiponectin acts through AdipoR1 and AdipoR2 in brain to promote neurogenesis and modulate synaptic activity in animals (Song et al., 2015; Yau et al., 2014; Zhang et al., 2017), adiponectin-mediated signaling mechanisms have not been directly demonstrated. APPL1, first identified as the intracellular binding partner of AdipoR1 and AdipoR2 using a yeast hybrid strategy, directly binds to the intracellular domains of AdipoR1 and AdipoR2 via its C-terminal phosphotyrosine-binding and coiled coil domains to mediate adiponectin signaling, including activation of AMPK (Liu et al., 2002; Mao et al., 2006; Mitsuuchi et al., 1999). In support of these observations, we showed that exogenous human recombinant adiponectin increases the levels of APPL1-coupled AdipoR1 and AdipoR2 as well as activated AMPKα1/2 (pT^183/172AMPKα1/2). We also made a novel observation that T-cadherin is able to bind to APPL1. Although there is evidence that T-cadherin is present in human brain (Takeuchi et al., 2000) and potentially mediates various metabolic effects (Chung et al., 2011; Fava et al., 2011; Fujishima et al., 2017; Org et al., 2009), little is known of T-cadherin signaling. T-cadherin has also been shown to be a receptor for hexameric and high-molecular-weight forms of adiponectin, which are essential for revasculization (Hug et al., 2004; Parker-Duffen et al., 2013). Because adiponectin in CNS exists predominantly as the low molecular weight version (Kusminski et al., 2007), our novel finding that exogenous recombinant human adiponectin induces T-cadherin and APPL1 association suggests that T-cadherin can bind exogenous adiponectin and signal in the brain. However, its downstream effector system(s) remain unknown.

In CNS, leptin was shown to modulate synaptic plasticity (Harvey et al., 2006), neuronal excitability (Harvey, 2007; Moult et al., 2009), firing frequency of dopaminergic neurons in ventral tegmental area (Trinko et al., 2011) and dendritic morphology in hippocampus (O’Malley et al., 2007). These data support the notion that leptin, acting through its cognate receptors, regulates a wide range of brain functions. The leptin receptor is a member of the class I cytokine-receptor family, with six isoforms derived from alternative mRNA splicing and/or proteolytic processing (Frühbeck, 2006). While sharing common extracellular and transmembrane domains, a, c, d and f OB-R isoforms have relatively short cytoplasmic domains, referred to as “short” isoforms. With ~300 more amino acids in the intracellular domain than the short isoforms, OB-Rb is referred to as the “long” isoform. In our current study, we show both long and short forms are present in the human frontal cortex. Furthermore, previous reports indicate that binding of leptin to its receptor OB-R results in rapid activation of intracellular JAK2 that is associated with OB-Rb leading to tyrosine phosphorylation of downstream JAK2 and STAT3 in cells (Banks et al., 2000; Ghilardi and Skoda, 1997; Guo et al., 2008; Kita et al., 2003). In accord, we showed that leptin increases tyrosine phosphorylation of JAK2 (pY^1007/1008) and STAT3 (pY⁷⁰⁵). More intriguingly, we made another novel observation that leptin stimulation elicits the recruitment of JAK2 to both long and short forms of OB-R, although the level of recruitment is less for OB-Ra compared to OB-Rb. The fact that insulin similarly reduces the leptin-induced JAK2 recruitment to both forms of OB-R suggests that this is a physiological action of leptin, at least in human cortex.

In summary, we have demonstrated that under ex vivo oxygenated and normothermic conditions, the postmortem human brain within 10 hours PMI, remains responsive to insulin, adiponectin and leptin. This finding is consistent with various earlier reports by us and others that postmortem brains can withstand some ischemic insults and retain some functions and ability to signal (Hahn et al., 2006; Talbot et al., 2012; Verwer et al., 2002; Vrselja et al., 2019; Wang et al., 2012; Wang and Friedman, 1994; Wang et al., 2003; Wang et al., 2017). Although these findings indicate that postmortem brains may be used to extract functional data and assess the impacts of disease, there are many limitations in studies using postmortem tissues. Most notably, molecules are vulnerable to postmortem changes and can lose integrity and functionality. However, we used tissues in as optimal condition as possible, by selecting tissue with postmortem intervals of less than 12 hours and without confounding diseases such as tumors or traumatic brain injury. We also used rigorous scientific methods including vigilant monitoring of storage temperature and detailed procedures for tissue preservation so that the integrity of the receptors and signaling molecules minimally compromised. Despite its limitations, the present work illuminated crosstalk and signaling details between leptin, adiponectin and insulin receptor signaling in the human brain. The present findings are an important first step in interrogating human brain changes of leptin, adiponectin and insulin receptor signaling pathways, as well as their crosstalk, in aging and age-related diseases such as ADRD. The loci of dysregulation that we identified in these fundamental metabolic pathways in the brain offer critical insights into pathophysiologic mechanisms of age-related neurologic diseases. These points of metabolic dysfunction may lead to new interventions so urgently needed for ADRD and other age-related diseases of the brain.

Highlights.

• Insulin, adipokine signaling, and their crosstalk can be measured in postmortem human brain.

• Adipokine enhances but leptin inhibits insulin signaling.

• Insulin reduces adiponectin but potentiates leptin signaling

Abbreviations:

AD

Alzheimer’s disease

AdipoR1

adiponectin receptor 1

AdipoR2

adiponectin receptor 2

ADRD

Alzheimer’s Disease-Related Dementias

AMPK

adenosine monophosphate dependent protein kinase

APPL1

adaptor protein containing pleckstrin homology domain phosphotyrosine-binding domain and leucine zipper motif

CSF

Cerebrospinal fluid

DMSO

dimethyl sulfoxide

EDTA

ethylenediaminetetraacetic acid

EGTA

aminopolycarboxylic acid

IR

insulin receptor

IRS-1

insulin receptor substrate 1

JAK2

Janus kinase 2

OB-R

leptin receptor

p38MAPK

p38 mitogen-activated protein kinase

PAGE

polyacrylamide gel electrophoresis

PBS

phosphate buffered saline

PMI

postmortem interval

PPAR-α

peroxisome proliferator-activated receptor-α

SDS

sodium dodecyl sulfate

STAT

signal transducer and activator of transcription 3

T2D

Type 2 Diabetes Mellitus