Various transgenic mouse lines to study proopiomelanocortin cells in the brain stem label disparate populations of GABAergic and glutamatergic neurons

Abstract

Products of the proopiomelanocortin (POMC) prohormone regulate aspects of analgesia, reward, and energy balance; thus, the neurons that produce POMC in the hypothalamus have received considerable attention. However, there are also cells in the nucleus of the solitary tract (NTS) that transcribe Pomc, although low levels of Pomc mRNA and relative lack of POMC peptide products in the adult mouse NTS have hindered the study of these cells. Therefore, studies of NTS POMC cells have largely relied on transgenic mouse lines. Here, we set out to determine the amino acid (AA) transmitter phenotype of NTS POMC neurons by using Pomc-Gfp transgenic mice to identify POMC cells. We found that cells expressing the green fluorescent protein (GFP) represent a mix of GABAergic and glutamatergic cells as indicated by Gad2 and vesicular Glut2 (vGlut2) mRNA expression, respectively. We then examined the AA phenotype of POMC cells labeled by a Pomc-Cre transgene and found that these are also a mix of GABAergic and glutamatergic cells. However, the NTS cells labeled by the Gfp- and Cre-containing transgenes represented distinct populations of cells in three different Pomc-Cre mouse lines. Consistent with previous work, we were unable to reliably detect Pomc mRNA in the NTS despite clear expression in the hypothalamus. Thus, it was not possible to determine which transgenic tool most accurately identifies NTS cells that may express Pomc or release POMC peptides, although the results indicate the transgenic tools for study of these NTS neurons can label disparate populations of cells with varied AA phenotypes.

INTRODUCTION

Neurons expressing the proopiomelanocortin (POMC) prohormone and its peptide products have been heavily studied because of the roles that POMC peptides play in reward, analgesia, and energy balance. POMC neurons are predominantly found in the arcuate nucleus (ARC) of the hypothalamus (25, 27, 44), and this population of POMC neurons clearly participates in energy balance regulation (32, 47). The high prevalence of obesity and desire to uncover the neural circuitry governing metabolic homeostasis have led to much investigation into ARC POMC cells. However, a second set of POMC neurons was identified by the immunohistochemical detection of the POMC peptides in the nucleus of the solitary tract (NTS) of young and colchicine-treated rats (7, 28, 36, 41). Evidence that metabolically relevant neurons reside in this brain region includes the observation that the NTS is innervated by vagal nerve afferents signaling nutritional information from the alimentary canal (1, 2, 6, 33). Furthermore, the use of Pomc promoter elements to drive the expression of green fluorescent protein (GFP; see Ref. 11) led to the discovery that Pomc-Gfp neurons in the NTS express leptin receptors (20) and respond to peripheral administration of leptin (15) as indicated by phosphorylation of signal transducer and activator of transcription 3 (STAT3). Additionally, c-FOS immunoreactivity is observed in NTS Pomc-Gfp cells following systemic administration of a meal-terminating dose of the gut peptide cholecystokinin (2, 17). Together, it appears that NTS POMC neurons respond to energy state. However, it is important to note that NTS neurons identified by their expression of Cre recombinase driven by Pomc genomic sequences (Pomc-Cre) do not show c-FOS immunoreactivity or STAT3 phosphorylation following leptin administration (24), questioning the leptin responsiveness of NTS POMC neurons.

In contrast to the ARC (43), detection of Pomc mRNA or peptides in adult mice within the NTS has proven difficult (39). Because of this, and the importance of Pomc neurons in metabolically relevant behaviors and the cellular heterogeneity of the brain regions in which they reside, recent studies of NTS POMC neurons have relied on transgenic mouse lines to study and manipulate these cells. To advance our understanding of NTS neurons expressing Pomc promoter-driven constructs, we designed the present experiments to determine if Pomc-Gfp and Pomc-Cre cells in the NTS, like those in the ARC, are a mix of GABAergic and glutamatergic cells (22, 26, 46). With the use of in situ hybridization, the present data show that NTS cells labeled with both Pomc-Gfp and Pomc-Cre transgenic approaches represent a mix of GABAergic and glutamatergic phenotypes, indicating the need to consider the contribution of these transmitters in mediating effects when these neurons are activated. In addition to the mixed amino acid phenotypes, we also found Pomc-Gfp and Pomc-Cre cells to be distinct nonoverlapping populations of cells. Using three separate Pomc-Cre driver lines, we show that Cre-expressing cells are located caudal and lateral to Pomc-Gfp cells in the NTS. These latter findings are in agreement with previous work (35) and may help clarify disparate findings reported in the literature where experiments were conducted in Pomc-Gfp- vs. Pomc-Cre-expressing cells.

MATERIALS AND METHODS

Animals.

The following mice were procured from Jackson Laboratories (Bar Harbor, ME): Pomc-Gfp [C57BL/6J-Tg(Pomc-EGFP)1Low/J, stock: 009593], ROSA-tom [129S6-Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze}/J, stock: 007905], and Pomc-Cre^Lowl [Tg(POMC1-Cre)16Lowl/J, stock: 005965]. Pomc-Cre^GSB mice (47) were originally a gift from Dr. Greg Barsh (Stanford University). Pomc-Cre:ERT2 (5) were originally a gift from Dr. Joel Elmquist (University of Texas Southwestern Medical Center). All lines were congenic to the C57BL/6J strain. Transgenes and floxed alleles were detected using standard polymerase chain reaction techniques. Male and female mice were used for all experiments, and data from both sexes were pooled for final analysis and data presentation, since no sex differences were observed. Animals were housed on a 12:12-h light-dark schedule and had ad libitum access to standard rodent chow and tap water. All animal use procedures were approved by the Colorado State University Institutional Animal Care and Use Committee and met United States Public Health guidelines.

Fluorescent in situ hybridization.

For tissue collection, mice were anesthetized with pentobarbital sodium followed by transcardiac perfusion of 10% sucrose in water followed by 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). Brains were then removed from the skull and stored overnight in 4% PFA at 4°C. Horizontal sections (50 μm) containing the NTS were cut on a vibratome in ice-cold diethylpyrocarbonate (DEPC)-treated PBS and processed for fluorescent in situ hybridization as detailed below. Brain slices were incubated in 6% H₂O₂ for 15 min, washed two times for 5 min each in PBS containing 0.1% Tween 20 (PBT), and treated for 15 min with proteinase K (10 μg/ml) in PBT. Proteinase K was then deactivated by exposing the brain slices to glycine (2 mg/ml) in PBT for 10 min. Tissue sections were postfixed for 20 min in a 4% PFA/0.2% gluteraldehyde solution, washed two times in PBT for 5 min, and then dehydrated in ascending concentrations of ethanol (50, 70, 95, and 100%) prepared in DEPC-treated water. Prehybridization of the tissue occurred for 60 min at 60°C in hybridization solution [66% (vol/vol) deionized formamide, 13% (wt/vol) dextran sulfate, 260 mM NaCl, 1.3× Denhardt’s solution, 13 mM Tris·HCl (pH 8.0), and 1.3 mM EDTA (pH 8.0)]. Probes were denatured for 5 min at 85°C and then added to the hybridization solution along with tRNA (0.5 mg/ml) and dithiothreitol (10 mM). Hybridization of the tissue occurred for 18–20 h at either 70 or 52°C for vesicular Glut2 (vGlut2)-DIG- and Gad2-FITC-labeled probes, respectively. We previously determined concentration and hybridization temperature for each probe used based on characteristic staining that was restricted to appropriate brain regions (13, 26, 27). The probe sequences and details of probe preparation have been previously described in full (26).

After hybridization was completed, brain slices underwent three 30-min stringency washes in 50% formamide/5× saline sodium citrate at 60°C followed by three 30-min washes in 50% formamide/2× saline sodium citrate at 60°C. Tissue was then digested with RNase A [20 μg/ml in 0.5 M NaCl, 10 mM Tris·HCl (pH 8.0), 1 mM EDTA] at 37°C for 30 min before being washed in TNT [0.1 M Tris·HCl (pH 7.5), 0.15 M NaCl, and 0.05% Tween 20] three times for 15 min each and then blocked in TNB (TNT plus 0.5% Blocking Reagent provided with the TSA kit; Perkin Elmer, Oak Brook, IL) for 1 h. Overnight incubation of tissue then took place at 4°C in either sheep anti-FITC antibody (1:1,000; Roche Applied Sciences) or sheep anti-DIG antibody (1:1,000; Roche Applied Sciences), both conjugated to horseradish peroxidase (HRP). DIG-labeled probes were detected using a TSA PLUS Biotin kit (Perkin Elmer) and subsequently incubated in 1% H₂O₂ to quench remaining peroxidase activity. To detect FITC-labeled probes, a TSA PLUS DNP (HRP) kit (Perkin Elmer) was used. Slices were washed in TNT for three 20-min washes followed by 30 min of exposure to either the Biotin Amplification Reagent (1:50) or the DNP Amplification Reagent (1:50). Tissue was washed again in TNT, and visualization of DIG-labeled probes was accomplished by exposing slices to streptavidin conjugated to Alexa Fluor 555 in TNT (1:1,000; Invitrogen, Eugene, OR) for 30 min. FITC-labeled probes were visualized by exposing slices to rabbit anti-DNP-KLH conjugated to Alexa Fluor 488 in TNT (1:400; Invitrogen) for 1 h. Slices were then mounted and coverslipped using Aqua Poly/Mount (Polysciences, Warrington, PA).

Dual fluorescent in situ hybridization and immunodetection of GFP.

In some instances, dual in situ hybridization using both FITC- and DIG-labeled probes was performed using the methods described above but with the following modifications. Gad2-FITC and vGlut2-DIG probes hybridize at different temperatures (52 and 70°C, respectively), so we performed sequential hybridization. The probe that hybridized at the higher temperature (70°C) was hybridized first for 18–20 h before being transferred to a new vial containing the second probe, and the tissue was hybridized a second time at 52°C for 18–20 h. After stringency washes were completed, DIG-labeled probes were detected first by using the TSA PLUS Biotin kit as described above. Tissue was then exposed to 1% H₂O₂ diluted in TNT for 1 h, washed three times for 20 min in TNT, and incubated overnight at 4°C in TNB containing the anti-FITC antibody (1:1,000). Slices were then washed in TNT, and FITC-labeled probes were visualized as described above using the TSA PLUS DNP kit.

Pomc-Gfp transgenic animals were used to determine if mRNA transcribing Gad2 or vGlut2 colocalizes with Pomc-Gfp cells. GFP fluorescence is quenched during in situ hybridization; however, antigenicity of GFP is maintained. GFP was visualized by incubating tissue for 2 h in TNT containing a chicken anti-GFP primary antibody (1:2,000; Abcam, Boston, MA). Slices were then exposed to a donkey anti-chicken secondary antibody conjugated to Alexa Fluor 647 (1:1,000; Jackson ImmunoResearch, West Grove, PA) for 1 h. Controls for in situ hybridization included use of sense probes in place of antisense sequences and omission of primary antibodies as detailed previously (26). Controls for the GFP antibody included use of GFP-negative tissue, omission of primary antibody, and accurate detection in the arcuate nucleus.

Imaging and analysis.

Images were acquired on either a Zeiss 510 or 880 laser-scanning confocal microscope. Fluorophores were imaged sequentially to avoid crossover between channels. Cell counts were made using a modification of a three-dimensional counting method (45). For each tissue section, 3-μm slices in the z-plane were collected in each channel. For analysis, each z-stack was pared down to five sequential slices, omitting slices taken at the surface of the tissue. GFP or enhanced yellow fluorescent protein (eYFP)-immunoreactive (IR) cells were identified using masks created automatically in National Institutes of Health ImageJ software to detect Alexa 647-labeled fluors. Only cells completely within a 300 × 300 × 2 μm counting box on the x–y–z planes were counted and analyzed for colocalization with Gad2 or VGlut2 probes. Average florescence intensity for Gad2 and vGlut2 probes had to be >10% above background for a cell to be considered positive. Quantitative results are expressed as means ± SE.

Stereotaxic surgery for viral-mediated gene delivery.

Mice (6–8 wk old) were induced into a deep anesthetic plane with isoflurane and were placed in a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA) fitted with a nose cone for delivery of isoflurane for the entirety of the operation. An adeno-associated viral vector (AAV) allowing for Cre recombinase-dependent expression of eYFP [AAV9.EF1a.DIO.eYFP.WPRE.hGH; U. Penn Vector Core (University of Pennsylvania); 200 nl, 6.69e13 GC/ml] or a viral vector allowing for Cre recombinase-dependent expression of channelrhodopsin (ChR2) [AAV9.EF1.dflox.hChR2(H134R)-mCherry. WPRE.hGH; U. Penn Vector Core, University of Pennsylvania; 200 nl, 7.24e13 GC/ml] was delivered over the course of 60 s in each side of the NTS (from lambda: anterior/posterior: −2.48; medial/lateral ± 0.40; dorsal/ventral – 4.30) into Pomc-Cre^GSB mice. The needle was left in place for 300 s following the end of solution delivery. The injection of AAV-eYFP was used to express a reporter in Pomc-Cre cells that could be detected with immunohistological techniques following in situ hybridization. As with GFP, the fluorescence of YFP is lost during the in situ hybridization protocol, but the antigenicity remains. Brain slices from mice injected with AAV-eYFP were prepared for in situ hybridization 15–20 days after viral injection.

Inducible expression of Pomc-Cre.

Tamoxifen was purchased from Sigma-Aldrich and prepared in corn oil (Sigma-Aldrich) at a concentration of 20 mg/ml. Tamoxifen was administered in the intraperitoneal cavity (75 mg·kg⁻¹·day⁻¹) of PomcCre:ERT2 mice for five consecutive days beginning at postnatal day 42. Mice were euthanized for experiments 3–4 wk later.

RESULTS

Amino acid phenotype of Pomc-Gfp cells in the NTS.

POMC neurons in the ARC express transcriptional markers indicating that, in addition to their peptide phenotype, they can also be GABAergic or glutamatergic, or in some cases both (23, 26). Additionally, cultured ARC POMC neurons (22) and those in ex vivo hypothalamic slice preparations (14) are capable of releasing either GABA or glutamate. Because Pomc-driven constructs are also expressed in neurons in the NTS, we designed initial experiments to determine if NTS Pomc cells express a similar amino acid neurotransmitter phenotype as those in the ARC. To accomplish this, horizontal slices from the caudal brain stem containing the NTS were prepared from transgenic animals expressing GFP driven by the Pomc promoter (11). In situ hybridization was then conducted on these slices using probes against mRNA transcribing Gad2 and vGlut2. Within the NTS, 28.2 ± 4.6% of Pomc-Gfp cells colocalized with Gad2 mRNA (n = 7 mice; 443 GFP cells counted; Fig. 1A), 42.1 ± 2.6% with vGlut2 mRNA (n = 6 mice; 413 GFP-IR cells counted; Fig. 1B), and 15.6 ± 2.3% of Pomc-Gfp cells colocalized with both vGlut2 and Gad2 mRNAs (n = 6 mice; 426 GFP cells counted; Fig. 1C). Gad1 mRNA was detected in the caudal brain stem (data not shown) but did not colocalize with Pomc-Gfp cells. Collectively, these in situ hybridization studies indicate that Pomc-Gfp neurons in the NTS may be capable of releasing the amino acid (AA) transmitters GABA and glutamate.

Fig. 1.

Proopiomelanocortin (Pomc)-green fluorescent protein (Gfp) cells in the nucleus of the solitary tract (NTS) express GABAergic and glutamatergic phenotypic markers. A: representative images of NTS neurons showing immunodetection of GFP (left, green), GABAergic (Gad2 mRNA (middle, magenta), and a merge of the two images (right). B: representative images of NTS neurons showing immunodetection of GFP (left, green), glutamatergic (vGlut2) mRNA (middle, magenta), and a merge of the two images (right). C: representative images of NTS neurons showing immunodetection of GFP (left, green), Gad2 mRNA (middle left, blue), vGlut2 mRNA (middle right, magenta), and a merge of the three images (right). For all images, white arrows point to some of the colocalized cells; scale bars: 50 μm.

Pomc-Cre and Pomc-Gfp cells differ in anatomical localization.

Activating POMC neurons and attempting to detect AA transmitter release from these cells would require the use of Pomc-Cre transgenic mice. Thus, we sought to verify that Pomc-Gfp and Pomc-Cre transgenic constructs were expressed in the same cells in the NTS. An AAV enabling Cre-dependent expression of ChR2 tagged with a mCherry fluorophore (ChR2^mCherry) was stereotactically injected in the NTS of mice expressing Pomc-Gfp and also Cre recombinase driven by the Pomc promoter (referred to here as Pomc-Cre^GSB; see Ref. 48). After allowing 2–3 wk for viral expression of ChR2, horizontal slices containing the NTS were prepared and imaged. In these slices, it was apparent that ChR2^mCherry-positive cells did not colocalize with Pomc-Gfp cells (Fig. 2), confounding comparisons with in situ hybridization results.

Fig. 2.

Expression of Cre-dependent channelrhodopsin in proopiomelanocortin (Pomc)-Cre cells does not overlap with Pomc-green fluorescent protein (Gfp) cells in the nucleus of the solitary tract (NTS). Representative confocal image showing that Pomc-Gfp cells do no colocalize with cells expressing Cre recombinase-dependent ChR2^mCherry (magenta) in slices taken from mice expressing both Cre recombinase and Gfp driven by Pomc promoter sequences. An adeno-associated viral vector (AAV) encoding Cre recombinase-dependent ChR2^mCherry was injected in the NTS 17 days before perfusion. Rostral (R) and lateral (L) orientations are indicated. Scale bar: 50 μm.

To further visualize anatomical distribution of Pomc-Gfp and Pomc-Cre, we used the Cre recombinase reporter mouse line GtROSA26Sor-tdTOM (referred to here as ROSA-tom; see Ref. 30). This mouse line expresses a floxed STOP sequence upstream of the red florescent protein td-tomato. When these mice are bred with a mouse containing cell type-specific Cre expression, the STOP cassette is deleted in Cre-expressing cells, resulting in td-tomato fluorescence. Using this strategy, we compared the distribution of Pomc-Gfp and Pomc-Cre in horizontal sections containing the NTS from two separate commonly available Pomc-Cre lines by generating mice expressing all three transgenes (i.e., Pomc-Gfp, Pomc-Cre, and ROSA-tom). In both the Pomc-Cre^Lowl (4) and Pomc-Cre^GSB (47) lines, we observed almost complete separation between GFP and td-tomato-expressing cells (Fig. 3), indicating divergent cellular populations. From the Pomc-Cre^GSB line, 1.6 ± 0.9% of GFP cells were colabeled with td-tomato (n = 3 mice; 241 GFP cells counted, 246 ROSA-tom cells counted; Fig. 3A). Similarly, in the Pomc-Cre^Lowl line, 3.8 ± 1.7% of Pomc-Gfp cells coexpressed td-tomato (n = 5 mice; 786 GFP cells counted, 1,101 ROSA-tom cells counted; Fig. 3B). As indicated in the representative images, td-tomato fluorescence was consistently observed caudal and lateral to Pomc-Gfp cells. Although empirically determining if both Cre lines label the same population of cells is technically impossible, qualitative evaluations from these two lines indicated a seemingly indistinguishable expression pattern. Ultimately, these findings may help explain conflicting reports in the literature where incongruent findings were reported but different cell populations were likely studied because of the use of differing transgenic mouse lines (for example, the reports from Refs. 15 and 24). These data are in agreement with previous reports (35) and further underscore the differential nature of NTS Pomc-Gfp and Pomc-Cre cells when Cre is constitutively expressed.

Fig. 3.

Divergent localization of the nucleus of the solitary tract (NTS) proopiomelanocortin (Pomc)-green fluorescent protein (Gfp) and Pomc-Cre cells in two different Cre driver lines. A: representative images showing the anatomical localization of Pomc-Gfp neurons (top left, green) and Pomc-Cre^GSB;ROSA-tom neurons (2nd row left, magenta) in the mouse NTS. A merge and higher magnification of the two images is shown (3rd and 4th rows) to demonstrate the low level of colocalization. B: representative images showing the anatomical localization of Pomc-Gfp neurons (top right, green) and Pomc-Cre^Lowl;ROSA-tom neurons (2nd row right, magenta) in the mouse NTS. A merge and higher magnification of the two images is shown on bottom (3rd and 4th rows) to demonstrate the low level of colocalization. Rostral (R) and lateral (L) orientations are indicated. All scale bars: 50 μm.

To overcome the issue of recombination of floxed alleles in cells that show Pomc promoter activation during development, a transgenic mouse expressing Cre recombinase under control of the Pomc promoter that is selectively activated by tamoxifen has been developed (referred to here as Pomc-Cre:ERT2; see Ref. 5). This strategy allows for activation of Cre in Pomc cells at a given point in the mouse’s life, providing temporal control over when recombination occurs. To compare the expression pattern of Pomc-Gfp neurons and Pomc-Cre:ERT2 cells in the NTS, we generated mice coexpressing this inducible Cre, Pomc-Gfp, and ROSA-tom.

Mice expressing all three transgenes were administered tamoxifen for five consecutive days (75 mg/kg, 1 time/day ip) beginning at postnatal day 42. Between postnatal day 61 and 68 mice were euthanized, and horizontal sections through the NTS were collected to compare expression of GFP and td-tomato. Although the expression of td-tomato was low relative to the constitutive Pomc-Cre lines (Fig. 3 vs. Fig. 4A), a similar pattern emerged; Pomc-Cre:ERT2 cells as indicated by td-tomato fluorescence were located caudal and lateral to Pomc-Gfp cells (n = 5 mice; 580 total GFP cells counted, 78 ROSA-tom cells counted; 0 colabeled cells; Fig. 4A). In contrast, coexpression of GFP and td-tomato is reliably observed in the arcuate nucleus of mice with this genotype (Fig. 4B).

Fig. 4.

Divergent localization of proopiomelanocortin (Pomc)-green fluorescent protein (Gfp) and Pomc-Cre:ERT2 neurons in the nucleus of the solitary tract (NTS). A: representative images showing the anatomical localization of Pomc-Gfp neurons (top left, green) and Pomc-Cre:ERT2;ROSA-tom neurons (left, magenta images) in the mouse NTS. Merged image is shown (3rd row). B: representative images showing the anatomical localization of Pomc-Gfp neurons (top right, green) and Pomc-Cre:ERT2;ROSA-tom neurons (right, magenta images) in a coronal section of the mouse arcuate nucleus (ARC). A merged image is shown in the 3rd row. Higher-magnification images of the boxed regions from the 3rd row are shown below in the 4th row. Rostral/lateral (R and L) and dorsal/lateral (D and L) orientations are indicated in A and B, top, respectively. Scale bars: 50 μm.

Amino acid phenotype of Pomc-Cre cells in the NTS.

Pomc-Cre transgenic mice have been used to activate NTS Pomc neurons and to label these cells for studies using reporter lines. Thus, we sought to determine if the Pomc-Cre cells have an AA transmitter phenotype. A virus encoding a Cre-dependent eYFP construct was injected in the NTS of adult Pomc-Cre^GSB mice to express a fluorescent reporter in Pomc-Cre cells that would retain its antigenicity following fluorescent in situ hybridization. Following a 3-wk recovery period to allow for recombination, horizontal slices (50 μm) were cut from the caudal brain stem containing the NTS, and in situ hybridization to detect vGlut2 and Gad2 was conducted. In total, 714 YFP-IR cells were counted (n = 3 mice). Within the NTS, 25.0 ± 6.5% of YFP-IR cells expressed Gad2 mRNA (Fig. 5A) and 23.7 ± 1.7% expressed vGlut2 mRNA (Fig. 5B). Only 1.0 ± 0.3% of YFP-IR cells expressed both vGlut2 and Gad2 mRNA. Table 1 shows the expression of AA phenotypic markers in Pomc-Cre cells compared with that in Pomc-Gfp cells. No eYFP-IR cells were detected when the same virus was injected in the NTS of mice that did not express the Cre recombinase transgene (data not shown).

Fig. 5.

Proopiomelanocortin (Pomc)-Cre cells in the nucleus of the solitary tract (NTS) express GABAergic and glutamatergic phenotypic markers. A: representative images of NTS neurons showing immunodetection of Cre-dependent yellow fluorescent protein (YFP) that was virally expressed in Pomc-Cre cells (left, green), GABAergic (Gad2) mRNA (middle, magenta), and a merge of the two images (right). B: representative images of NTS neurons showing immunodetection of YFP (left, green), glutamatergic (vGlut2) mRNA (middle, magenta), and a merge of the two images (right). For all images, white arrows point to some of the colocalized cells; scale bars: 20 μm.

Table 1.

Percentage of Pomc cells in the NTS that express mRNA indicating a Gad2 and/or vGlut2 phenotype

	Pomc Cells with GABAergic or Glutamatergic Markers, %
	Pomc-Gfp	Pomc-CreGSB
Gad2	28.2 ± 4.6	25.0 ± 6.5
vGlut2	42.1 ± 2.6	23.7 ± 1.7
Gad2 and vGlut2	15.6 ± 2.3	1.0 ± 0.3

Values are means ± SE. Pomc, proopiomelanocortin; Gfp, green fluorescent protein; NTS, nucleus of the solitary tract; Pomc-Cre^GSB, Cre recombinase driven by the Pomc promoter; Gad2, GABAergic; vGlut2, glutamatergic.

DISCUSSION

Relevance of amino acid phenotype of Pomc-Gfp neurons in the NTS.

Hypothalamic POMC neurons are a critical component of the neural circuitry that governs energy homeostasis and food intake (10, 32). In addition to the release of peptides, POMC neurons in the ARC express transcriptional markers indicative of GABAergic and glutamatergic phenotypes (9, 13, 23, 27, 46), and individual POMC neurons can release either GABA or glutamate (14). Considering that the NTS population of putative POMC neurons has also been linked to the regulation of energy balance (40, 50) and other autonomic processes (8), we designed experiments to examine if NTS Pomc-Gfp cells also express mRNA for Gad2 or vGlut2.

Glutamate decarboxylase (GAD) is an enzyme necessary for the conversion of glutamate to GABA and, as such, Gad mRNA, is a reliable marker for GABAergic neurons. Using fluorescent in situ hybridization probes to detect the presence of Gad2 mRNA (encoding the GAD65 enzyme), the present data indicate that ~28% of Pomc-Gfp neurons in the NTS are GABAergic. Comparatively, Gad expression is found in ~35% of ARC POMC neurons (22). The GAD enzyme has two isoforms (16), and, although ARC POMC neurons express mRNA for both isoforms (26), we find only Gad2 mRNA in NTS Pomc-Gfp cells. The detection of Gad2 in brain-stem Pomc-Gfp neurons indicates that these cells may be capable of influencing target neurons though fast inhibitory modulation.

To determine whether Pomc-Gfp neurons in the NTS also have a glutamatergic phenotype, we performed in situ hybridization studies that revealed the presence of vGlut2 mRNA in ~42% of Pomc-Gfp neurons. This compares to ~7% of ARC POMC neurons (26). Because vesicular glutamate transporters (VGLUTs) are necessary to package glutamate into vesicles for release into presynaptic terminals (19), the detection of vGlut mRNA is a reliable marker for glutamatergic neurons. Therefore, the detection of vGlut2 mRNA in a subset of NTS Pomc-Gfp cells suggests that these neurons are capable of packaging and releasing glutamate.

The possibility that some NTS Pomc-Gfp neurons may have a dual amino acid transmitter phenotype was suggested by the presence of both vGlut2 and Gad2 mRNA in a subset (~15%) of Pomc-Gfp neurons. This compares to ~3% of ARC POMC cells expressing both vGlut2 and Gad mRNA (26). Dual amino acid phenotypes have been detected in multiple other brain regions (38, 42, 49), and convergent evidence suggests that many cells can undergo phenotypic plasticity to adapt to changing stimuli such as stress and fasting (12, 13, 27), although the significance of such a phenotype for NTS Pomc-Gfp neurons is unknown.

Divergent anatomical localization of NTS Pomc transgenic neurons.

We initially set out to investigate if NTS Pomc neurons are capable of releasing GABA/glutamate as indicated by the expression of Gad2 and vGlut2. However, studies of functional release require use of Pomc-Cre transgenic mice, and early experiments to verify that Pomc-Cre and Pomc-Gfp were expressed in the same NTS neurons did not reveal colocalization. Rather, the cells labeled by Pomc-Gfp and the cells expressing Pomc-Cre were distinct in their anatomical location. The clear anatomical separation between Pomc-Gfp and Pomc-Cre cells is consistent with a previous report (35) and held true for the Pomc-Cre^GSB, Pomc-Cre^Lowl, and Pomc-Cre:ERT2 lines.

It is tempting to attribute at least a portion of the expression of Cre-dependent reporter in non-Pomc-Gfp cells to the transient developmental expression of the Pomc promoter in non-POMC cells (34). However, the transient expression of Pomc during development was avoided in the experiments where we visualized Pomc-Cre cells by the expression of ChR2^mCherry after introducing the Cre-dependent viral construct in the NTS of adult mice. With this approach, the Pomc-Cre and Pomc-Gfp cells also appeared as separate populations as indicated by the lack of overlap between mCherry and GFP. Furthermore, use of an inducible Pomc-Cre (Pomc-cre:ERT2) mouse to drive Pomc-Cre only in adulthood (5) also showed the Pomc-Gfp cells and Pomc-Cre to be distinct populations. Therefore, developmental expression of Pomc does not explain the lack of overlap between the Pomc-Gfp and Pomc-Cre cells.

It is important to note that the Pomc-Cre and Pomc-Gfp constructs reliably and specifically label authentic POMC-peptide-expressing neurons in the ARC (4, 5, 11, 48), and, as such, the expression of Cre and GFP driven by the Pomc promoter is colocalized with POMC cells in the ARC, not in distinct sets of cells as in the NTS (also see Fig. 4). In many cells and brain regions, transgenic markers have been shown to accurately reflect endogenous mRNA expression patterns for a gene of interest. However, this congruency is not always observed (21, 29). The underlying nature of this disconnect is unclear. One possibility is that different approaches to produce the transgenes underlie the differing expression (e.g., inclusion or omission of distinct regulatory elements). The Pomc-Gfp construct is a traditional transgene containing 11 kb of upstream regulatory sequence, whereas each version of Cre expression used a bacteriological artificial chromosome (BAC) transgenic approach. It is tempting to speculate that the BAC approach may more accurately reflect endogenous expression patterns because of the likely inclusion of more extensive regulatory elements. However, it is important to note that the pattern of Pomc-Gfp expression more closely resembles the pattern noted in the few reports of POMC peptide and mRNA localization in the NTS (medial rostral). Furthermore, a recently developed Pomc-Cre knockin mouse did not produce evidence of Cre-mediated recombination in the NTS (18). Although it is not clear that transgenic approach alone dictates expression to distinct cells, each line will have distinct sites of integration and, thus, distinct regulation and expression strength.

Determining whether Pomc-Cre or Pomc-Gfp expression more accurately reflects true POMC neurons remains elusive. Although some studies have reported immunoreactivity for POMC peptides in slices containing the NTS from young colchicine-treated rats (7, 28, 31, 36) or from NTS extracts (37), many other studies (including our own) have failed to detect POMC peptides in the NTS despite clear labeling in the ARC. Indeed, the difficultly in detecting POMC peptides or Pomc mRNA in the NTS has been well documented (39), and we did not detect Pomc mRNA in the NTS in the present work. A few studies have detected Pomc mRNA in the NTS of mice using radioactive approaches with amplification (20, 35). Unfortunately, these approaches are not readily compatible with detection of the reporters as used in the present study. Nonetheless, previously reported localization of Pomc mRNA in the NTS exhibits a very medial distribution more in line with the pattern of Pomc-Gfp observed here compared with the somewhat more lateral distribution noted for the Cre reporter expression.

The lack of Cre in the NTS in the knockin POMC-Cre mice (18) combined with the dearth of evidence for POMC peptide expression in the mouse NTS indicate that these cells may be phenotypically distinct from POMC neurons as classically defined. Nevertheless, NTS cells identified by fluorescence driven by Pomc genomic sequences are consistently demonstrated to play important roles in autonomic processes and to respond to energy balance cues (2, 3, 8, 15, 17, 24, 40, 50).

Amino acid phenotype of Pomc-Cre neurons in the NTS.

In spite of the fact that Pomc-Gfp and Pomc-Cre cells were found to be distinct populations of cells in the present work, we decided to characterize the amino acid phenotype of the Pomc-Cre cells. We reasoned that AA phenotype may be another means by which to distinguish the two populations of cells. However, using Pomc-Cre to drive a fluorescent reporter and performing in situ hybridization to detect vGlut2 and Gad2 revealed that, like Pomc-Gfp neurons, Pomc-Cre neurons in the NTS are also a mix of glutamatergic and GABAergic cells. Thus, AA phenotype does not distinguish the two cell populations but does indicate an important consideration for using Pomc-Cre-based approaches to alter the activity of these cells, since changes in GABA and glutamate release will likely result and may contribute to any noted consequences. This is a particularly important point since the expression of optogenetic or chemogenetic tools in NTS Pomc-Cre cells has been used to evaluate the role of these neurons in cardiac, respiratory, analgesic, and metabolic responses (8, 50). Considering the lack of evidence indicating clear expression of POMC peptides in the NTS, our results suggest that behavioral phenotypes generated from chemogenetic or optogenetic stimulation of NTS Pomc-Cre are likely an effect of GABA or glutamatergic transmission. In addition, the finding that Cre and GFP are in separate cells when driven by the Pomc promoter suggests that studies aimed at mapping the outputs of NTS Pomc neurons could produce different results based on the transgenic mouse employed.

Perspectives and Significance

The present data indicate that both Pomc-Gfp and Pomc-Cre cells in the NTS express mixed amino acid transmitter phenotypes but that these cells represent separate populations. The expression of GABAergic and glutamatergic markers these cells suggests that Pomc neurons in the caudal brain stem are capable of influencing target neurons and brain regions through the rapid transient action of AA transmitters. The divergent anatomical localization of these two transgenic reagents indicates that comparisons between cells labeled with these two approaches are confounded. It remains to be determined which, if either, of these tools is expressed in authentic POMC peptide-expressing neurons in the NTS.

GRANTS

This work was supported by National Institutes of Health Grants DK-078749 and DA-032562 (S. T. Hentges) and the Monfort Family Foundation (S. T. Hentges).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.R.R., A.R.H., and S.T.H. conceived and designed research; A.R.R., A.R.H., and S.T.H. performed experiments; A.R.R. and A.R.H. analyzed data; A.R.R., A.R.H., and S.T.H. interpreted results of experiments; A.R.R. prepared figures; A.R.R. and S.T.H. drafted manuscript; A.R.R. and S.T.H. edited and revised manuscript; A.R.R., A.R.H., and S.T.H. approved final version of manuscript.