In vivo measurement of enhanced agouti-related peptide release in the paraventricular nucleus of the hypothalamus through G_s activation of agouti-related peptide neurons

Obesity is a burgeoning health crisis in most parts of the world [1]. It is a major risk factor for many diseases such as type 2 diabetes, cancer, and cardiovascular disease [2]. Weight loss and management are difficult to achieve via behavioral approaches, and only a limited number of appetite-suppressing drugs show modest clinical efficacy. Thus, efforts are being made to identify neuronal circuits that regulate food intake in order to more selectively target the development of new therapeutics [3,4].

There is a growing body of evidence that the neuropeptide agouti-related peptide (AgRP), which is expressed exclusively in the arcuate nucleus (ARC) of the hypothalamus, plays a critical role in controlling food intake, body weight and metabolism, and, most importantly, managing the long term stability of body weight [5-7]. For example, the firing rate of AgRP neurons is elevated in brain slices of food-deprived mice [8], and ablation of AgRP neurons in adult mice leads to starvation [9,10]. Further, the neurotransmitters AgRP and neuropeptide Y (NPY), a peptide that coexists with AgRP, increase food intake when injected into the brain [11,12]. Optogenetic and chemogenetic methods can mimic endogenous regulation of AgRP neurons [6,13-19]. Optogenetic stimulation of AgRP neurons triggers a robust increase of food intake in well-fed mice [20]. DREADDs (designer receptors exclusively activated by designer drugs) are designer G-protein coupled receptors (GPCRs) that can be activated by clozapine-N-oxide (CNO), an otherwise pharmacologically inert drug [21-24]. DREADD technology has been used to activate AgRP neurons through the G_q- and G_s-mediated signaling pathway [6,19] which mimic the activity of endogenous ghrelin and PAC1 receptors at AgRP neurons [25]. Similar to light activation, Gq or Gs DREADD stimulation of AgRP neurons leads to robust food intake, even during the light cycle when food intake is typically restricted [6,19].

It remains unclear how GPCRs signaling in AgRP neurons in the ARC affects the release of AgRP in brain regions monosynaptically connected to and down-stream from these neurons, such as the paraventricular nucleus (PVN) of the hypothalamus. This is due to the challenges of developing an assay to monitor the in vivo kinetics of AgRP, a large molecule (112 amino acids). Thus, the purpose of this study was to determine whether excitation of AgRP neurons in the ARC by DREADD technology triggers AgRP release in the PVN. Microdialysis is a sampling technique which has been extensively used for the characterization and assessment of the in vivo pharmacokinetic profile of drugs and neurotransmitters in rodents and non-human primates. Recently, with the availability of larger pore membrane probes and the push-pull method, in vivo microdialysis has been used to detect relatively large peptides in the brain (e.g., apolipoprotein E [26]; tau protein [27]; amyloid β [28]). We took advantage of this technique and investigated AgRP release in the PVN after DREADD-mediated activation of the G_s pathway in AgRP neurons. Here we report the successful monitoring of AgRP level in the mouse PVN via microdialysis.

All experiments were conducted according to US National Institutes of Health guidelines for animal research and were approved by the NIDDK Institutional Animal Care and Use Committee. Twenty-four AgRP-ires-cre mice (Agrptm1(cre)Lowl/J; The Jackson Laboratory, stock #012899) were randomly divided into two groups and were injected bilaterally with either an adeno-associated virus encoding G_s–DREADD (AAV8-hSyn-DIO-Rs-eGFP (AV4557, titer 8 × 10¹²)) or a control virus (AAV8-hSyn-DIO-GFP (AV4937, titer 2 × 10¹²) into the ARC. Both AAVs were obtained from the UNC Vector Core Facility. Microdialysis probes were placed in the PVN during a second surgery 6 weeks after the viral injections.

Male mice (at least 8 weeks old) were anaesthetized with isoflurane and placed into a stereotaxic apparatus (David Kopf Instruments, model 940A with 923B mouse gas anesthesia head holder). The skull was exposed via a small incision, and holes were drilled (0.45 mm drill bits) into the skull. A 5 μl Hamilton syringe with a 33 gauge, blunt-end needle was inserted into the brain, and bilateral pressure injections of the AAV were administered targeting the ARC (Bregma: −1.46 mm AP; ± 0.3 mm ML; −5.80 mm DV). The injection volume was 200–300 nl depending on the titers of different viruses. Postoperative analgesia was provided (ketoprofen, 5 mg/kg i.p.).

The microdialysis probe used for in vitro study had an 8 mm shaft with a 1 mm, 1000 kDa molecular weight cutoff (MWCO) polyethylene (PE) membrane (PEP-8-01, Eicom, Kyoto, Japan). The probe was prepared by flushing it first with distilled water and then with artificial cerebrospinal fluid (aCSF) perfusion solution. The probe’s inlet and outlet were then connected to a microsyringe pump (kdScientific, KDS-250) and a peristaltic pump (LongerPump, BT100-2J with DG4-10 head), respectively, using polyethylene tubing (PE-20, 380 μm ID). Sampling was carried out at different flow rates (0.5–2.0 μl/min). Probes were placed into AgRP solutions of known concentrations (60 and 600 pg/ml), and samples were collected every 50 min and stored at −80°C until being processed.

Mice were anaesthetized using 2.5% tribromoethanol (Avertin^®) and placed into a stereotaxic apparatus. Before the probes were implanted, a small incision was made to expose the skull, the head position was leveled using bregma and lambda as reference points, and a small hole was drilled for injections and implants. Probes were connected to the perfusion system as described for the in vitro microdialysis experiment. The pump speeds were calibrated such that the fluid perfused through the probes at a 1 μl/min flow rate. Finally, the probe was implanted in the PVN (bregma: −0.75 mm AP; 0.13 mm ML; −5.60 mm DV) and fixed using an anchoring bone screw and dental cement.

Mice were given 24 h to recover, then dialysate samples were collected every 50 min for about 2 h to establish an equilibrium between the inside and outside of the microdialysis membrane and acclimate the mice to the collection procedure (Fig. 1). Baseline samples were collected over 50 min. Mice then received a CNO injection (1 mg/kg i.p.), and the perfusion solution was allowed to flow through the probe for 30 min before the post-CNO samples were collected, at 1 and 24 h. Samples were collected into vials containing 5 µl 0.1 N HCl and immediately frozen on dry ice. All samples were stored at −80°C until processed, and all stereotaxic sites were histologically verified.

Dialysate samples from 10 mice were collected, and AgRP concentrations were measured in these samples using a mouse enzyme immunoassay kit (US Biological, 023199). However, prior to the assay, one sample from two mice, matched by condition (i.e., baseline or 1 h- or 24 h-post CNO injection), were randomly paired and pooled to yield the required sample volume and detectable concentrations for the ELISA kit. The detection limit of the enzyme immunoassay kit was 4.69 pg/ml with a sensitively of 1.83 pg/ml. The intra-assay CV was 1.44% and 3.70%, respectively.

To inhibit axonal transport and enhance our ability to visualize AgRP by immunohistochemical (IHC) staining, 30 µg of colchicine in 5 µl of physiological saline was delivered under anesthesia at a flow rate of 500 ng/min intracerebroventricularly (bregma: −0.34 mm AP; −1.0 mm ML; −3.0 mm DV). The mice were returned to their home cage and left undisturbed for 24 h. Then, mice were perfused with 4% paraformaldehyde in 0.1 M phosphate buffer fixative (pH 7.4), and brains were harvested and postfixed in this solution overnight. Mouse brains were sliced with a vibratome (Leica VT1000S) at 30 µm thickness and stored in 4°C in PBS until assayed. For IHC staining, brain slices were incubated overnight at 4°C with primary antibodies (either anti-AgRP, 1:1000, H-003-57, Phoenix Pharmaceuticals, INC or anti-GFP, 1:400, ab6662(FITC)) diluted in PBS, supplemented with 1% BSA and 0.1% Triton X-100. Slices were then washed three times and incubated with species appropriate and minimally cross-reactive fluorophore-conjugated secondary (Abcam; Donkey anti rabbit IgG 594, 1:000, ab150064, Abcam; goat anti rabbit 594, 1:1000) for 2 h at room temperature. Slices were rinsed in PBS with 0.1% Triton X-100 three times and mounted for imaging using Antifade Reagent (SlowFade^® Gold with DAPI, S36939). Fluorescence images were taken with Zeiss and Leica fluorescent microscopes.

Fresh frozen mouse brains were collected and stored at −80°C, then 16 µm sections were cut with a cryostat and mounted onto superfrost slides and stored at −80°C. The 2-plex 2.0 kit was used for RNAScope assays. Following the manufacturer’s (Advanced Cell Diagnostics) protocol, sections on the slides were fixed at room temperature with 10% NFB for 30 min, dried with 50%, 70% and 100% ethanol. The edges of the slides were marked with a waterproof pen. Pretreatment steps included room temperature incubations with pretreatment solution 1 (10 min), pretreatment solution 2 (10 min), and pretreatment solution 3 (30 min). Sections were hybridized at 40°C for 2 h with a specifically designed AgRP probe. Finally, sections were exposed to serial amplification steps (AMP1 through AMP6), color development, counterstaining with 10% hematoxylin, and dried by using a 65°C heat block. Cover slides were placed over the sections on the slide with Ecomount, and images were captured and processed with a Keyence Digital Microscope.

Statistical analyses were performed using IBM SPSS Statistics 19 (SPSS, Inc., an IBM Company) and were expressed as mean ± SEM. Normality was tested using Kolmogorov-Smirnov. Outliers were assessed using the IQR (interquartile range) method, denoting values three times below the 1^st quartile or above the 3^rd quartile as outliers. No sample calculation was performed.

We injected G_s–DREADD or AAV-hSyn-DIO-GFP viruses into male AgRP-ires-cre mice and eight weeks later visualized the expression using IHC to visualize viral GFP expression. GFP expression was localized to cell bodies in the ARC and colocalized with AgRP expression, confirming that viral expression was selective to AgRP neurons (Fig. 2). AgRP protein and mRNA expression were quantified to determine if viral administration affected these parameters. However, neither G_s–DREADD nor AAV-hSyn-DIO-GFP administration influenced AgRP protein or mRNA levels, visualized by IHC and RNAScope, respectively (Fig. 3).

We selected a microdialysis probe membrane (1000 kDa MWCO, PEP-8-01, Eicom, Kyoto, Japan) suitable for collecting high-molecular weight molecules, as AgRP has a molecular mass of ~14 kDa. There was an inverse relationship between flow rate and percentage recovery for AgRP at a low and a high concentration, 60 and 600 pg/ml respectively (flow rate: 0.5 μl/min, 26.0% recovery; 1.0 μl/min, 17.6% recovery; 2.0 μl/min, 6.5% recovery) (Fig. 4). Based on these results, we decided to use the 1.0 μl/min flow rate for in vivo microdialysis experiments (see next paragraph).

To determine if G_s activation increased AgRP release, we used chemogenetic technology and real-time in vivo microdialysis to stimulate AgRP neurons in the ARC and measure subsequent AgRP release in the PVN. Specifically, we bilaterally injected the G_s-DREADD virus into the ARC of AgRP-ires-Cre mice to express the G_s-linked DREADD in AgRP ARC neurons (Fig. 2). Eight weeks later, the extracellular fluid from the PVN was collected to measure the AgRP concentrations before and after (1 h- and 24 h-post) CNO injections. Extracellular AgRP concentrations were measured 94.07 ± 2.65 pg/ml in the saline-treated mice and 90.71 ± 6.93 pg/ml in the CNO-treated mice. CNO-mediated excitation of AgRP ARC neurons markedly enhanced the release of AgRP in the PVN during the first hour (t(4) = 5.52, P < 0.005) following the CNO injection but not 24 h later (t(4) = 0.29, P = 0.79) (Fig. 4). Control mice received either the G_s-DREADD virus without CNO treatment (saline + DREADD; virus control) or were treated with CNO in the absence of G_s-DREADD expression (CNO + AAV-hSyn-DIO-GFP; drug control). No effects on AgRP release were observed for either control group.

In this study, we used in vivo microdialysis with a unique large-pore-sized membrane probe (1000-kDa MWCO) and push-pull perfusion system to monitor AgRP extracellular levels in the PVN, a major target region of hypothalamic AgRP neurons. To the best of our knowledge, this is the first report that has demonstrated the detection and characterization of AgRP release in the brain, which is important to determine the function of AgRP signaling in regulation of food intake, metabolism, and body weight.

Microdialysis is a widely used technique to measure the release of neurotransmitters and other small molecules in the brain [29]. Low MWCO microdialysis probes (typically 6–18 kDa) prohibit collection of relatively large-sized molecules with efficiency, including peptides near the MWCO such as AgRP (~14 kDa MW). Several recent reports have demonstrated that the use of high MWCO microdialysis probes improves collection of larger molecules and peptides (e.g., apolipoprotein E [26]; tau protein [27]; amyloid β [28]). Here, we first characterized the efficiency of AgRP collection from aCSF using microdialysis in vitro. The recovery of AgRP by microdialysis was inversely related to the flow rate through the microdialysis probe, which is consistent with measurements of other analytes [29]. We determined that a 1.0 µl/min flow rate yields the best recovery rate within the largest sample volume.

Our in vivo microdialysis work confirmed that activation of G_s signaling in AgRP neurons activation increased release of AgRP in the PVN. Specifically, we were able to selectively express a Gs DREADD in AgRP neurons in the ARC, which was confirmed to also express in axonal fibers in the PVN. Using site-specific injections of CNO into the PVN, the AgRP concentration in the PVN was increased an hour later. PVN AgRP levels returned to baseline within 24 h after stimulation. Recent studies showed that CNO can convert in vivo into clozapine, which has binding affinity and potency to targets other than DREADD [30]. Thus, to exclude non-specific outcomes of CNO metabolite action, we administered CNO into the PVN of DREADD-free mice, which did not affect local AgRP concentrations. Previous studies indicate that AgRP affects hunger-triggered food intake, stability of body weight, and long-term energy balance [5-7], which may occur, at least in part, through signaling via G_s-linked GPCRs on AgRP neurons in the ARC [19]. There are several brain regions downstream of the ARC AgRP neurons that have garnered attention regarding food intake research, such as PVN, bed nucleus of the stria terminalis and lateral hypothalamus. The methods developed here will be useful in sampling AgRP in the extracellular fluid of these brain regions over time in order to assess the relationship between AgRP levels and feeding behaviors and the role of G_s-linked GPCRs in mediating these processes.

Several studies have shown that AgRP neuron activation alone is sufficient to trigger robust food intake [6,19,20], which has led to the characterization of these neurons as ‘hunger neurons’ [7,31,32]. Still, how precisely the activation of AgRP neurons initiates feeding is still under investigation. There is evidence that gamma-Aminobutyric acid (GABA) release from these neurons is a rapid response to inhibit downstream neurons to trigger robust feeding acutely. In contrast, Krashes and colleagues demonstrated that the AgRP-induced food intake is impaired by double inactivation of both the GABA and NPY systems, but the effect only lasted only for a few hours [31]. This suggests that AgRP release may play an important role in long term food intake. In fact, it has recently been reported that G_s pathway activation, with the same DREADD system used in the current study, promotes food intake that lasts for days and mimics the effects of i.c.v. AgRP administration [19]. In contrast, stimulation of Gq pathway with a similar DREADD system (i.e., CNO-treated hM3Dq-AgRP mice) lasted only for a single day. This may result from the difference in cellular activity caused by the activation of Gs- versus Gq-dependent signaling pathways, which may also be reflected in distinct patterns of AgRP release. The effect of G_s activation could be blocked by a selective anti-AgRP antibody. In this study, we clearly demonstrated in vivo that G_s activation increases the release of AgRP in the PVN. Thus, enhancing AgRP release or G_s signaling in AgRP neurons may represent a useful strategy to develop novel appetite-suppressing drugs.

Special thanks to Drs. Oksana Gavrilova, Hans-Jurgen Wess, and W. Scott Young as well as June Song for technical assistance with the RNAScope technique and supplying the AgRP-ires-cre mice. In addition, thanks to Drs. Gavrilova, Wess, and Young for reviewing an early version of this manuscript. This research was supported by the intramural research programs of the NIMH (ZIA-MH-002498-24) and NIDDK (DK070002).

Scientific concept, experimental design, and data analysis: Cui Z and Smith AS. Viral injections, immunohistochemistry, and imaging: Cui Z. In vitro and in vivo microdialysis and AgRP ELISA: Smith AS. We thank Drs. Oksana Gavrilova and Hans-Jurgen Wess for providing AgRP-ires-cre mice. We also thank Dr. W. Scott Young and June Song for performing the RNAScope to detect AgRP mRNA on tissue sections.