Subregional Differences in GABA_A Receptor Subunit Expression in the Rostral Ventrolateral Medulla of Sedentary Versus Physically Active Rats

Abstract

Neurons in the rostral ventrolateral medulla (RVLM) regulate blood pressure through direct projections to spinal sympathetic preganglionic neurons. Only some RVLM neurons are active under resting conditions due to significant, tonic inhibition by gamma-aminobutyric acid (GABA). Withdrawal of GABA_A receptor-mediated inhibition of the RVLM increases sympathetic outflow and blood pressure substantially, providing a mechanism by which the RVLM could contribute chronically to cardiovascular disease (CVD). Here, we tested the hypothesis that sedentary conditions, a major risk factor for CVD, increase GABA_A receptors in RVLM, including its rostral extension (RVLM_RE), both of which contain bulbospinal catecholamine (C1) and non-C1 neurons. We examined GABA_A receptor subunits GABA_Aα1 and GABA_Aα2 in the RVLM/RVLM_RE of sedentary or physically active (10–12 weeks of wheel running) rats. Western blot analyses indicated that sedentary rats had lower expression of GABA_Aα1 and GABA_Aα2 subunits in RVLM but only GABA_Aα2 was lower in the RVLM_RE of sedentary rats. Sedentary rats had significantly reduced expression of the chloride transporter, KCC2, suggesting less effective GABA-mediated inhibition compared to active rats. Retrograde tracing plus triple-label immunofluorescence identified fewer bulbospinal non-C1 neurons immunoreactive for GABA_Aα1 but a higher percentage of bulbospinal C1 neurons immunoreactive for GABA_Aα1 in sedentary animals. Sedentary conditions did not significantly affect the number of bulbospinal C1 or non-C1 neurons immunoreactive for GABA_Aα2. These results suggest a complex interplay between GABA_A receptor expression by spinally-projecting C1 and non-C1 neurons and sedentary versus physically active conditions. They also provide plausible mechanisms for both enhanced sympathoexcitatory and sympathoinhibitory responses following sedentary conditions.

GRAPHICAL ABSTRACT

Sedentary conditions produce greater dendritic branching in bulbospinal catecholamine (C1) neurons associated with greater sympathoexcitation to direct glutamatergic activation of the rostral ventrolateral medulla (RVLM). We identified a greater percentage of bulbospinal C1 neurons but fewer absolute number of bulbospinal non-C1 neurons expressing GABA_Aα1 receptor subunits in sedentary versus physically active rats. Complex expression of GABA_A receptors on different subpopulations of bulbospinal neurons may explain enhanced sympathoexcitatory as well as sympathoinhibitory responses following sedentary conditions.

1. Introduction

The rostral ventrolateral medulla (RVLM) is one of the most important brain regions involved in control of resting blood pressure (Dampney et al., 2003a; Guyenet, 2006; Schreihofer and Sved, 2011). In addition, there is strong evidence that overactivity of neurons in the RVLM contributes to cardiovascular diseases such as hypertension and heart failure (Sved et al., 2003; Wang et al., 2009; Huber and Schreihofer, 2011). More recently, structural and functional changes in RVLM neurons have been associated with sympathetic overactivity in sedentary versus physically active rats (Mischel et al., 2015; Mueller et al., 2017). Given the association between functional and structural neuroplasticity in the RVLM and their connection to a variety of risk factors for cardiovascular disease (Mischel et al., 2015; Mueller et al., 2017), understanding alterations in the primary neurotransmitter systems that regulate the activity of RVLM neurons continues to be of critical importance.

Although a subset of RVLM neurons are active at rest and serve to maintain normal blood pressure, a potentially larger proportion of RVLM neurons have either very low activity or are entirely silent due to significant inhibition by gamma-aminobutyric acid (GABA) (Dampney et al., 2003a; Guyenet, 2006; Schreihofer and Sved, 2011). For example, several investigators have provided functional evidence of prominent, tonic inhibition in the RVLM, reporting that blockade of GABA_A receptors elicits substantial increases in sympathetic outflow and blood pressure in anesthetized rats (Horiuchi et al., 2004; Mueller, 2007) and rabbits (Horiuchi and Dampney, 1998). Inhibition of the RVLM is mediated in part by input from arterial baroreceptors via connections with GABAergic neurons in the caudal ventrolateral medulla (Schreihofer and Guyenet, 2003; Guyenet, 2006; Mueller, 2007). However, a substantially larger fraction appears to come from non-baroreceptor-related inputs, as evidenced by blockade of GABA_A receptors producing much larger increases in sympathetic nerve activity than decreases in blood pressure in experiments that directly compared these responses (Mueller et al., 2011). Thus, GABA receptors, particularly of the GABA_A subtype, are critically important in the moment-to-moment regulation of sympathetic outflow and blood pressure (Dampney et al., 2003a; Guyenet, 2006; Schreihofer and Sved, 2011).

Alterations in the number or distribution of RVLM neurons expressing GABA receptors may play significant roles in disease states associated with sympathetic overactivity. Similarly, other factors that regulate GABAergic inhibition, such as the K⁺/Cl⁻ co-transporter (KCC2), can also contribute to sympathoexcitatory states such as after high salt intake (Choe et al., 2015). However, to our knowledge, no studies have tested the mechanisms by which sedentary conditions influence GABAergic inhibition directly by quantifying changes in expression of GABA receptors or KCC2 in the RVLM.

Our laboratory has provided functional evidence for alterations in GABAergic transmission in the RVLM of sedentary versus physically active rats (Mueller, 2007; Mueller and Mischel, 2012; Dombrowski and Mueller, 2017). For example, sedentary animals exhibit greater sympathoexcitatory responses to blockade of GABA_A receptors in the RVLM (Mueller, 2007), enhanced sympathoinhibitory responses to GABA microinjections in the RVLM (Dombrowski and Mueller, 2017), and greater GABAergic modulation of pressor responses produced by direct activation of RVLM neurons with glutamate (Mueller and Mischel, 2012). Interestingly, these data suggest that sedentary conditions may upregulate GABA receptors, a proposition that is counterintuitive because sedentary conditions are associated with sympathetic overactivity (Mueller et al., 2017). However, from a functional standpoint, withdrawal of increased GABAergic tone would be consistent with enhanced sympathoexcitatory responses to decreases in blood pressure, which have been reported previously in sedentary versus physically active animals (DiCarlo and Bishop, 1988; Negrao et al., 1993; Mischel and Mueller, 2011). Therefore, the purpose of this study was to determine whether sedentary versus physically active conditions result in differences in the expression and/or distribution of proteins involved in GABAergic signaling within the RVLM (i.e. GABA_A receptor subunits and KCC2). Since sedentary conditions are associated with both an increase in sympathoexcitation in response to withdrawal of GABAergic inhibition of the RVLM and enhanced sympathoinhibitory responses to GABA, we tested the hypothesis that sedentary conditions result in an increase in the number of bulbospinal RVLM neurons expressing GABA receptors, which would be consistent with the augmented sympathetic nerve responses (both sympathoexcitation and sympathoinhibition) observed in vivo.

2. Methods

2.1. Ethical statement

The Wayne State University Institutional Animal Care and Use Committee approved all procedures and experiments. We also followed the National Academy of Sciences’ Guiding Principles in the Care and Use of Animals as adopted by the American Physiological Society. For both ethical and experimental design purposes, we performed all experiments with an effort to maximize the amount of brain tissue used from each animal. In cases where the amount of remaining tissue was not sufficient for further analysis (e.g., KCC2 western blotting), animals were added to both sedentary and active groups (see protocols below for details). Using this approach, we were able to make between-group comparisons for the effects of sedentary versus active conditions and use a repeated-measures analysis for examination of rostrocaudal effects within every animal.

2.2. Animal models

As described previously, we studied male Sprague Dawley rats (n=55; Harlan/Invigo, Indianapolis, IN), which were purchased at four weeks of age and weighed 75–99 g. Animals were housed in a temperature- and light- controlled facility (12:12 light:dark cycle), which was AALAC-accredited (Animal Welfare Assurance Number A3310–01). Upon arrival, each rat was randomly placed in a standard polycarbonate cage either with a commercially available running wheel (Techniplast, Eaton,PA; physically active group) or without a wheel (sedentary group). Licensed animal care staff checked animals daily for food (Purina LabDiet 5001: Purina Mills, Richmond, IN) and tap water, which were provided ad libitum. For active rats, laboratory personnel documented daily running distances and durations, as recorded by bicycle computers (Sigma Sport, Olney, IL) affixed to each cage. Animals were maintained under sedentary or active conditions for 10–12 weeks.

2.3. Validation of antibodies

Based on commercial availability of reagents and our experimental design, we chose to examine two GABA receptor subunits (GABA_Aα1 and GABA_Aα2) in western blotting and immunofluorescence studies and KCC2, a chloride ion transporter that has been shown to indirectly influence chloride conductance through GABA receptors (Sigel and Steinmann, 2012), in an additional series of western blotting studies. We also used GAPDH as a loading control in western blotting experiments. Tables 1 and 2 contain the details regarding sources of antisera, the immunogens employed, and dilutions used.

Table 1 -.

Primary Antibodies

Antibody	Immunogen	Source, Cat. #, Species	RRID	Dilution
GABAAα1	Amino acids 28-43 of rat GABAAα1	Alomone Labs, AGA-001, rabbit polyclonal	AB_2039862	1:500 (IF) 1:400 (WB)
GABAAα2	Amino acids 395-405 of rat GABAAα2	Alomone Labs, AGA-002, rabbit polyclonal	AB_2039864	1:200 (WB)
GABAAα2 clone N399/19 supernatant	Amino acids 350-385 of rat GABAAα2	NeuroMab, 73-384, mouse monoclonal	AB_2336904	1:5 (IF) 1:2.5 (WB)
KCC2	Amino acids 932-1043 of rat KCC2	EMD Millipore, 07-432, rabbit polyclonal	AB_310611	1:500 (WB)
TH	TH from rat pheochromocytoma	EMD Millipore, AB152, rabbit polyclonal	AB_390204	1:1000 (IF)
TH Clone LNC1	TH purified from PC12 cells	EMD Millipore, MAB318, mouse monoclonal	AB_2201528	1:500 (IF)
CTB	CTB purified from Vibrio cholerae	List Biological, 703, goat polyclonal	AB_10013220	1:10 000 (IF) 1:200 000 (IP)
GAPDH Clone 6C5	GAPDH from rabbit muscle	EMD Millipore, MAB374, mouse monoclonal	AB_2107445	1:1000 (WB)
Actin	C-terminus of human actin	Santa Cruz, Sc-1616, goat polyclonal	AB_630836	1:250 (WB)

GABA, gamma amino-butyric acid; KCC2, potassium chloride cotransporter 2; TH, tyrosine hydroxylase; CTB, cholera Toxin Beta subunit; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase; IF, Immunofluorescence; IP, Immunoperoxidase; WB, western blotting.

Table 2 -.

Secondary antibodies

Antibody	Immunogen	Source, Cat. #,	RRID	Dilution
Goat anti-Rabbit, HRP conjugated	Whole rabbit IgG	Millipore, 12-348, polyclonal	AB_11214240	1:5 000 (WB)
Goat anti-Mouse HRP conjugated	Whole mouse IgG	Millipore, 12-349, polyclonal	AB_390192	1:5 000 (WB)
Donkey anti-goat HRP conjugated	Whole goat IgG	Santa Cruz, Sc-2020, polyclonal	AB_631728	1:15 000 (WB)
Donkey anti-Goat Biotin conjugated	Whole goat IgG	Jackson ImmunoReaserch, 705-065-147, polyclonal	AB_2340397	1:500 (IF, IP)
Donkey anti-Mouse Biotin conjugated	Whole mouse IgG	Jackson ImmunoReaserch, 715-065-151, polyclonal	AB_2340785	1:500 (IF)
Donkey anti-Rabbit Biotin conjugated	Whole rabbit IgG	Jackson ImmunoReaserch, 711-065-152, polyclonal	AB_2340593	1:500 (IF)
Donkey anti-Rabbit Alexa Fluor 594 conjugated	Whole rabbit IgG	Invitrogen, A-21207, polyclonal	AB_141637	1:500 (IF)
Donkey anti-Goat Alexa Fluor 594 conjugated	Whole goat IgG	Life Technologies, A-11058, polyclonal	AB_2534105	1:500 (IF)
Donkey anti-Goat Alexa Fluor 647 conjugated	Whole goat IgG	Invitrogen, A-21447, polyclonal	AB_2535864	1:200 (IF)
Donkey anti-Mouse Alexa Fluor 594 conjugated	Whole mouse IgG	Life Technologies, A-21203, polyclonal	AB_2535789	1:500 (IF)

All other primary antibodies used for immunofluorescence studies had been characterized by the manufacturer or validated in our previous reports (Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014). Several (except for those against GABA_A receptors) are also listed in the Journal of Comparative Neurology antibody database. The TH and CTB antibodies used in our previous publications (Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014) produced identical staining patterns in the present experiments. When secondary antisera were used alone, no staining was observed.

2.4. Western blotting

2.4.1. Tissue preparation

Rats from physically active and sedentary groups (n=17) were anesthetized with Fatal Plus (0.25 mL/kg, i.p.; Vortech, Dearborn, MI; 390 mg/ml of sodium pentobarbital). A loss of muscle tone and lack of response to firm pinch were confirmed before decapitation with a rodent guillotine. The whole brain was quickly removed from the skull and divided into brainstem and forebrain using a rodent brain matrix (Braintree Scientific, Braintree, MA). Forebrains and brainstems were frozen quickly on a stainless steel metal plate over dry ice and stored in a −80 °C freezer until further cryosectioning.

Brainstems were cut into 80 μm coronal sections using a Microm HM 550 cryostat (ThermoScientific, Waltham, MA; see Figure 1), and the sections were placed on uncoated, precleaned microscope slides. We collected bilateral micropunches along the rostrocaudal extent of the medulla using a 17gauge blunt needle with a 1 mm internal diameter. We obtained punches with the aid of a rat atlas (Paxinos and Watson, 2007) and using as landmarks the lateral edge of the pyramidal tract, the medial and ventral edge of the spinothalamic tract and the area ventral to the nucleus ambiguus. Based on studies from our laboratories and others, this area contains bulbospinal C1 and non-C1 neurons (Burman et al., 2004; Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014). Each set of bilateral punches was placed in 20μl of ice-cold lysis buffer containing 150 mM NaCl, 1 mM EDTA, 50 mM Tris, 1% IGEPAL CA-630, 0.1% Triton X-100, and Halt protease inhibitor cocktail (Pierce Biotechnology, Rockford, IL), which was used at a 1× working concentration (1mM AEBSF, 800nM Aprotinin, 50μM Bestatin, 15μM E64, 20μM Leupeptin, and 10μM Pepstatin A).

Figure 1. Preparation of sections of the ventrolateral medulla for western blotting.

Diagram showing serial sectioning of hindbrain tissue for western blotting analysis of the rostral ventrolateral medulla (RVLM) and its rostral extension (RVLM_RE). Brainstem sections were cut at 80 μm and bilateral tissue punches were obtained and pooled based on their location relative to the caudal pole of the facial nucleus (designated FN0 and depicted by the solid vertical line). The RVLM and the RVLM_RE were each divided into two regions based on their locations caudal and rostral to FN0, respectively (RVLM: FN-240 and FN-480; RVLM_RE: FN+240 and FN+480; see Material and Methods). Regions FN-240, FN-480, FN+240 and FN+480 each contained three sets of bilateral punches (designated by the grey dashed lines). NAc = compact division of the nucleus ambiguus.

After micropunches had been taken, sections were fixed on slides in 4% formaldehyde for 30 min, washed three times with phosphate buffer and stained with cresyl violet to verify the rostral and caudal boundaries of the RVLM (Pawar et al., 2017). Similar to our previous immunohistochemistry studies (Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014), sections were aligned serially and their positions identified relative to the caudal pole of the facial nucleus (FN0). Based on this alignment, micropunches from corresponding RVLM locations were pooled by combining bilateral punches from each of three serial 80 μm-thick sections, representing 240 μm subregions of the RVLM per pooled sample (~1mg tissue). Because the RVLM is traditionally defined as the 600 μm caudal to the caudal pole of FN, which equates to FN0 (Dampney et al., 2003a; Guyenet, 2006; Schreihofer and Sved, 2011), samples were pooled into subregions using the caudal pole of FN as the defining landmark. We and others have reported the presence of bulbospinal C1 and non-C1 neurons rostral to FN0 (Dampney et al., 2003a; Guyenet, 2006; Schreihofer and Sved, 2011; Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014), a region that we define here as the rostral extension of the RVLM or RVLM_RE. Therefore, two subregions of the traditional RVLM were obtained by pooling three 80 μm punches i.e. FN-80 to FN-240 (designated as FN-240) and FN-320 to FN-480 (designated as FN-480). Similarly, two subregions of the RVLM_RE were obtained by pooling three 80 μm punches i.e. FN+80 to FN+240 (designated as FN+240) and FN+320 to FN+480 (designated as FN+480) (Figure 1).

Samples were vortexed briefly and then sonicated twice at 45% amplitude for three seconds using an ultrasonic processor (Fisher Scientific, Waltham, MA). In order to avoid overheating, samples were placed on ice and cold water was used in the sonicator cuvette. After protein extraction, samples were centrifuged in a cold room at 12,000 rpm for 15 min and supernatants were collected in fresh sample tubes for protein analysis. Protein concentrations were determined using a bicinchoninic (BCA) protein assay kit (Pierce Biotechnology, Rockford, IL) and measured at 562 nm using a Synergy H1 hybrid plate reader (Biotek, Winooski, VT).

We also performed preliminary testing on two loading controls, actin and GAPDH. Only GAPDH was expressed at consistent levels regardless of experimental group or rostrocaudal subregion of the RVLM. Overall, actin also showed significantly less intense bands compared to GAPDH. Furthermore, actin was consistently expressed at higher levels in sedentary versus physically active animals and demonstrated an inconsistent signal intensity along the rostrocaudal extent of the RVLM and RVLM_RE (data not shown). Thus, GAPDH was used as a loading-control protein throughout all blotting experiments and all data are presented as the ratio between the protein of interest (GABA_Aα1, GABA_Aα2, and KCC2) and GAPDH.

2.4.2. Immunoblotting

Five microgram samples of protein homogenate from each subregion (−480; −240; +240; +480) from each animal were separated at 110 volts on 10% (for GABA_A receptors; n=12 for each group) or 7.5% (for KCC2; n=9 for each group), polyacrylamide gels using Tris/Glycine/SDS running buffer (Bio-Rad, Richmond, CA). For studies on both GABA receptor subunits, we used homogenates from the same set of 12 animals in each group. For KCC2 studies, we used the remaining homogenate from four sedentary and four active rats of the original 12 in each group, plus new homogenates from an additional five sedentary and five active rats in order to provide appropriate sample sizes.

After electrophoretic separation, proteins were transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad, Richmond, CA) at 400 mA for two hrs in a cold room. Following transfer, the membranes were incubated in blocking solution containing 5% nonfat dry milk (Bio-Rad, Richmond, CA) and 3% normal goat serum (NGS) in 1xPBS, pH 7.4 with 0.1% Tween 20 (1xPBST) for 1 hr at room temperature on a shaker. Membranes were then incubated overnight at 4°C in one of five different primary antibodies (rabbit anti-GABA_Aα1; rabbit anti-GABA_Aα2; mouse anti-GABA_Aα2; rabbit anti-KCC2; or mouse anti-GAPDH) diluted in 1xPBST with 2.5% dry milk and 1% NGS. After three 10 min washes, membranes were incubated for 1 hr at room temperature in a species-specific secondary antibody conjugated to horseradish peroxidase (HRP, see Table 2), diluted in 1xPBST with 2.5% dry milk and 1% NGS. Following washes in PBST, protein bands were detected using enhanced chemiluminescent HRP substrate (ECL Western Blotting Detection Reagents, GE Healthcare, Piscataway, NJ) and film autoradiography (Blue Lite radiography films; GeneMate/BioExpress, Kaysville, UT). Developed films were scanned at high resolution and all signals were quantified using densitometric analysis and Image J software (NIH, Bethesda, Maryland).

2.5. Immunofluorescence studies of bulbospinal C1 and non-C1 neurons

2.5.1. Spinal cord injections of retrograde tracer

As in our previous studies (Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014), all rats (11 sedentary, 10 active) received spinal cord injections of cholera toxin B subunit (CTB; List Biologicals, Campbell, CA) seven days prior to sacrifice. Briefly, isoflurane-anesthetized rats were placed in a stereotaxic frame and 1% CTB was injected bilaterally into the intermediolateral cell column (IML) at spinal cord segments T9–T10 (three 60 nl injections on each side). Upon return to sternal recumbency, animals were returned to their home cage with clean bedding for one week of recovery. Running wheels were provided to physically active rats after spinal cord surgery, since our previous studies have already demonstrated a consistent albeit lower level of voluntary activity after surgery. Wheels remained in the cages of active rats until the day before sacrifice, at which time they were removed to minimize any potential acute effects of exercise (Kajekar et al., 2002).

2.5.2. Tissue preparation

On the day of sacrifice, rats were deeply anesthetized with Fatal Plus (0.25 mL/kg, i.p.) and were perfused transcardially with 500 ml of oxygenated DMEM/F12 tissue culture medium (D-8900; Sigma Chemical, St. Louis, MO), followed by 1 L of freshly prepared 4% formaldehyde in 0.1 M phosphate buffer, pH 7.4 (Llewellyn-Smith and Mueller, 2013; Mischel et al., 2014). Brains and spinal cords (with the vertebral column intact), were removed and post-fixed in 4% formaldehyde solution for an additional 2–3 days. Spinal cords were removed carefully from their vertebral columns and post-fixed in 4% formaldehyde for an additional 2–3 days. Tissues were rinsed repeatedly with phosphate buffer after post-fixation.

Blocks containing the medullas of physically active or sedentary rats (n=10 per group) were prepared with the aid of a rodent brain matrix (Braintree Scientific) and stored in PB containing 0.05% sodium azide to inhibit bacterial growth until cut. Brainstem blocks from physically active or sedentary rats (n=10 per group) were cryoprotected in 20% sucrose in 10 mM Tris, 0.9% NaCl, 0.05% thimerosal in 10 mM phosphate buffer, pH 7.4 (TPBS), followed by 30% sucrose in TPBS for at least 24 hours at each sucrose concentration, and stored in −80°C freezer until cryosectioning. The right side of each medulla was marked with a pinhole and medullas were sectioned in the coronal plane at 30 μm on a cryostat (HM550 VP, Waldorf Germany, Thermo Scientific). Sections were collected serially into four separate wells containing PB, transferred into cryoprotectant solution and stored at −20°C until processed for immunofluorescence.

Following extensive washing in PB, spinal cords from each animal were prepared for single immunofluorescence staining or single immunoperoxidase staining. For single immunofluorescence staining, spinal cords were divided into 12 thoracic segments according to dorsal root entry zones and each segment was placed in separate wells in a 12-well plate for sucrose cryoprotection. The segments were then embedded in two Tissue-Tek cryomolds (EMS, Hatfield, PA) in tissue freezing medium OCT (EMS, Hatfield, PA) with six spinal cord segments per mold. Blocks were stored in −20°C freezer until sectioning. Blocks containing the T9–T10 spinal cord segments were cryosectioned transversely at 20 μm and sections were collected on Histo-band slides (Fisher Scientific, Pittsburgh, PA) or gel-coated slides and stored in −80°C freezer until processed for single immunofluorescent staining. For single immunoperoxidase staining, spinal cord segments were embedded in albumin-gelatin followed by sucrose cryoprotection (Llewellyn-Smith et al., 2007, Llewellyn-Smith and Mueller, 2013, Mischel et al., 2014). The embedded T9–T10 spinal cord segments were cryosectioned serially at 20 μm and placed sequentially into six wells containing PB to produce six sets of sections at 120 μm intervals. Sections were transferred into cryoprotectant and stored at −20°C until processed for immunoperoxidase staining according to our previous protocol (Llewellyn-Smith et al., 2005; Llewellyn-Smith and Mueller, 2013).

2.5.3. Triple Label Immunofluorescent Staining - Brainstem sections

Two sets of sections from the 1:4 series of sections cut from each rat medulla were removed from cryoprotectant solution and placed in a Petri dish containing PB. The RVLM/RVLM_RE-containing sections from each rat were selected using the anatomical landmarks from a rat atlas (Paxinos and Watson, 2007) and a dissecting microscope. Typically, 12 medulla sections from each animal were processed for triple label immunofluorescence according to our established protocol (Llewellyn-Smith and Mueller, 2013). Briefly, sections were washed 3 × 15 mins in TPBS with 0.3% Triton X-100 (TPBS/Triton) followed by preincubation in blocking solution containing 10% normal horse serum (NHS; Life Technologies, Grand Island, NY) in TPBS/Triton for 30 mins. Sections were incubated in a mixture of primary antibodies diluted in TPBS/Triton containing 10% NHS for three days at room temperature. CTB plus TH plus the GABA_Aα1 receptor subunit were localized in one set of sections from nine sedentary and eight active rats using goat anti-CTB, mouse anti-TH and rabbit anti-GABA_Aα1. In a second set of sections from eight of the same sedentary and eight of the same active rats, plus sections from two additional sedentary and two additional active rats (n=10 total in each group), CTB plus TH, plus the GABA_Aα2 receptor subunit were localized using goat anti-CTB plus rabbit anti-TH plus mouse anti-GABA_Aα2 (Table 1).

After 3 × 30 min washes in TPBS, sections were incubated overnight at room temperature in a mixture of species-specific secondary antibodies conjugated to Alexa 594 (red fluorescence, TH) and Alexa 647 (fluorescence represented as blue, CTB) in TPBS/Triton containing 1% NHS (Table 2). Immunoreactivity for GABA_Aα1 or GABA_Aα2 subunit was detected using a biotinylated donkey anti-rabbit or biotinylated donkey anti-mouse secondary antibody in 1% NHS-TPBS followed by overnight incubation with streptavidin conjugated to Alexa 488 diluted 1;1000 in TPBS/Triton (green fluorescence). Incubations were performed in the dark to minimize bleaching of fluorophores. After extensive washing in TPBS, sections were arranged serially in rostrocaudal order using a dissecting microscope and mounted on slides using Vectashield Mounting Medium (Vector Laboratories). Slides were coverslipped, sealed with nail polish, and stored in the dark at 4°C until microscopic examination.

2.5.4. Validation of CTB injection sites

Spinal cord sections were processed to identify the location of CTB injection sites via either immunofluorescence and immunoperoxidase staining techniques, each briefly described below.

Single immunofluorescence staining.

Slides containing the T9–T10 spinal cord sections were removed from the freezer, placed in a Coplin staining jar and washed 3×15 min in TPBS followed by a pre-incubation in 10% NHS in TPBS/Triton. Sections were incubated overnight in primary antibody (1:10,000, goat anti-CTB; Table 1) diluted in TPBS/Triton with 10% NHS. After extensive washing, slides were incubated in secondary antibody (1:500 donkey anti-goat immunoglobulin (IgG) conjugated to Alexa 594 or donkey anti-goat IgG conjugated to Alexa 647; Table 2) in the dark for two hours at room temperature. Immunolabeled slides were washed in TPBS, coverslipped using Vectashield (Vector Laboratories, Burlingame, CA) and stored at 4°C until confocal microscopic examination.

Single immunoperoxidase staining.

Sections from one well per rat were incubated in 1:200,000, goat anti-CTB (Table 1) followed by incubation in 1:500 biotinylated donkey anti-goat IgG (Table 2) and lastly, incubation in 1:1500 ExtrAvidin-peroxidase (Sigma, St. Louis, MO). CTB-Immunoreactivity at spinal cord injection sites was revealed by a peroxidase reaction with an imidazole-intensified diaminobenzidine - glucose oxidase reaction (Llewellyn-Smith et al., 2005). Stained sections were mounted on gel coated slides, dehydrated and coverslipped with Permount mounting medium (Fisher Scientific, Pittsburgh, PA) for light microscopic examination.

2.6. Data collection and analysis

Prior to confocal microscopy, brightfield images were taken of all 12 medullary sections from each rat to verify their positions along the rostrocaudal axis of the RVLM. Based on the identification of the caudal pole of FN in each rat, ten sections were selected for further confocal microscopic examination, including five sections caudal to the caudal pole of FN, the section containing the caudal pole (FN0) and the four sections rostral to the caudal pole of FN. As a result, the 1:4 series of sections spanned the RVLM from −600 μm caudal to +480 μm rostral to the caudal pole of FN with 90 μm between sections. The sections for analysis were designated FN-600, FN-480, FN-360, FN-240, FN-120, FN0, FN+120, FN+240, FN+360 and FN+480. We defined sections FN-600 to FN-120 as comprising the RVLM and sections FN0 to FN+480 as comprising the RVLM_RE. Due to the extensive cell counts that we performed and the lack of evidence to our knowledge of a priori differences in sidedness, only the right hemisphere of each section was used for microscopic examination and cell counting.

Images were captured on a laser scanning confocal microscope (LSM-780, Zeiss, Thornwood, NY) using a 40×/1.30 oil DIC M27 objective lens. Each fluorophore used in our experiments was excited with laser-specific wavelengths: Alexa 488 (Ex_max=498 nm, Em_max=520 nm) with 488 nm argon laser; Alexa 594 (Ex_max=589 nm, Em_max=612 nm) with 561 nm HeNe laser; and Alexa 647 (Ex_max=650 nm, Em_max=667 nm) with 633 nm HeNe laser. Sequential scanning was performed to eliminate crosstalk between channels and to separate signals from each other. In order to examine the RVLM at each rostrocaudal level, tiled images were captured at 512 × 512 pixels per tile (212.55 μm × 212.55 μm,) with line and frame average set up to four. The tiled area completely contained all C1 (TH+) neurons seen at each RVLM level.

We focused our studies on bulbospinal neurons retrogradely labeled by CTB from T9–T10, which eliminated the need to determine the presence or absence of GABA_Aα1 or GABA_Aα2 receptor subunit immunofluorescence on the larger population of non-bulbospinal cells in the region. Because all of the neurons examined in this study contained CTB and the majority contained TH, we were also able to use the morphology of each cell body to determine whether or not each neuron contained GABA receptor subunit immunofluorescence. The use of laser confocal microscopy allowed us to examine 1 μm thick optical sections on channels optimized to the excitation and detection range of each fluorophore. Bulbospinal (CTB+), C1 (CTB+/TH+) and non-C1 (CTB+/TH-) neurons immunoreactive for GABA_Aα1 or GABA_Aα2 were counted on digital micrographs at each rostrocaudal level using ImageJ software (NIH, Bethesda, Maryland). The triple-labeled image was opened in one window as a colorized, overlaid image. Using the Point Tool function in Add to Overlay mode, the blue CTB+ cells were marked with dots such that each positively-identified CTB-labeled cell was automatically identified on the other fluorophore channels (i.e. red = TH positive and green = GABA_A receptor subunit positive). Each CTB+ cell was then carefully examined for TH-immunoreactivity and GABA receptor subunit-immunoreactivity by examining the single channel for each of the other fluorophores.

Section-by-section rostrocaudal comparisons.

Counts of single (CTB+/TH-), double (CTB+/TH+ or CTB+/TH-/GABA subunit+), and triple-labeled neurons (CTB+/TH+/GABA subunit+) were reported as mean + standard error of the mean at each rostrocaudal level for each rat at 120 μm intervals from FN-600 through FN+480 (total 10 sections/animal). For counts comparing the rostrocaudal location of bulbospinal C1 and non-C1 neurons, we analyzed the expression of the GABA_Aα1 and GABA_Aα2 subunits separately, using two way mixed analysis of variance (ANOVA; see details below). We compared sedentary versus physically active animals (main effect of group) and the rostrocaudal location of counts performed on a section-by-section basis (main effect of FN number). Also because sedentary conditions resulted in a significantly lower absolute number of bulbospinal non-C1 neurons (see below), we also calculated the number of bulbospinal C1 and non-C1 neurons that were immunopositive for GABA_A receptor subunits as a percentage of the total number of bulbospinal C1 or bulbospinal non-C1 neurons at each rostrocaudal level from FN-600 through FN+480 using the same two way mixed ANOVAs for group and FN number.

RVLM versus RVLM_RE comparisons.

Bulbospinal C1 and non-C1 neurons are distributed differentially across the ventral portion of the medulla; that is, C1 neurons outnumber non-C1 neurons in the traditional RVLM; whereas, both cell types are fairly common in the rostral extension, RVLM_RE (Schreihofer and Sved, 2011). We have previously observed increased dendritic branching in bulbospinal C1 neurons from sedentary rats in a striking pattern showing more branching in rostral versus caudal C1 neurons (see Mischel et al., 2014), i.e. the equivalent of the RVLM_RE and RVLM regions, respectively, which we distinguish in the present study. In addition to increased dendritic branching, it is possible that inactivity-related neuroplasticity in the ventrolateral medulla involves changes in the numbers of bulbospinal neurons that express GABA_A receptor subunits and that these changes vary along the rostrocaudal axis of the RVLM_RE versus RVLM. Therefore, we examined the influence of sedentary versus physically active conditions on the total number of bulbospinal C1 and non-C1 neurons expressing GABA receptor subunits in the RVLM versus the RVLM_RE. For these analyses, we used two way mixed ANOVAs to compare sedentary versus physically active conditions (main effect of group) and the RVLM versus the RVLM_RE (main effect of subregion).

Bulbospinal C1 versus bulbospinal non-C1 comparisons. Because our experimental design allowed us to localize GABA_Aα1 and GABA_Aα2 subunits in the second and fourth series of sections from a subset of the same rats (n=8 per group), we were able to combine the cell counts from both series for each animal and produce the equivalent of counting half the total number of bulbospinal C1 or non-C1 neurons in the right RVLM/RVLM_RE from each rat. Combining these data provided more statistical power and less variability in assessing the total number of bulbospinal C1 and non-C1 neurons in sedentary versus physically active rats overall and in the RVLM versus the RVLM_RE. We used two way mixed ANOVA to compare between sedentary and physically active conditions (main effect of group) and between the RVLM and RVLM_RE (main effect of subregion).

We performed statistical analyses with SigmaStat Version 3.5 (Systat Software, San Jose, CA). We used unpaired t-tests to compare between groups of sedentary and active rats: organ and body weights and the percentage of bulbospinal neurons that were of the C1 subtype or the non-C1 subtype. We also used two-way mixed ANOVA to determine differences between active and sedentary groups (non-repeated measures) and differences in subregional expression (repeated measures in each animal). When we encountered significant main effects (in the absence of a significant interaction term), we performed post hoc multiple comparisons to determine differences within significant main effects (multiple comparisons adjusted by Holm-Sidak method). When we observed a significant interaction term, we performed post hoc, pairwise multiple comparisons tests (also adjusted by Holm-Sidak method) which allowed us to interpret differences in sedentary versus physically active animals in individual subregions including: 1) GABA receptor subunit or KCC2 expression in specific subregions of the ventrolateral medulla (Results Section 3.3); 2) numbers and percentages of C1 or non-C1 neurons expressing GABA receptor subunits at particular levels of the medulla (Results Sections 3.4.1 and 3.4.2); 3) the numbers or percentages of C1 or non-C1 neurons expressing each of the two GABA receptor subunits examined in the RVLM compared to the RVLM_RE (Results Sections 3.4.1 and 3.4.2); or 4) the total numbers and percentages of bulbospinal C1 or non-C1 neurons using combined counts from two sets of tissue from a subset of the same animals in each group, and independent of GABA receptor subunit staining (Results Section 3.4.3). We confirmed that all data sets were normally distributed and passed equal variance by using the original data set; the original data set following transformation (e.g., square root, log 10, etc. with or without prior reflection); or ranking, prior performing the two-way mixed ANOVA. We report all data as the average + standard error of the mean. We set a p value of less than 0.05 as the maximum threshold for denoting statistical significance and report p values under 0.1 to allow the reader to interpret our findings more fully. We list exact p values above 0.001 and p values below 0.001 as p<0.001 in the Results section and list both p value results and F statistics in each figure legend or in separate tables where necessary (see Appendices).

3. Results

3.1. Animal model

As expected from our previous work (Mischel and Mueller, 2011; Subramanian and Mueller, 2016; Dombrowski and Mueller, 2017), sedentary rats had significantly higher body weights but significantly lower right ventricle and adrenal gland weights compared to active rats. Soleus muscle and left ventricular weights were not different between groups (Table 3).

Table 3 -.

Body and organ weights and running distances.

	Left ventricle	Right ventricle	Left adrenal	Right adrenal	Soleus muscle	Body weight (g)	Total distance (km)
	(mg/g#) n=17					n=27	n=27
Sedentary	2.11±0.03	0.47±0.01	0.05±0.003	0.05±0.001	0.37±0.01	402±6	N/A
Active	2.16±0.03	0.51±0.01*	0.06±0.002*	0.06±0.002*	0.38±0.01	357±8*	183±25

Body weights are presented in grams and organ weight expressed as mg tissue/g body weight. All data expressed as mean ± s.e.m.

The total distance that active rats ran over the 10–12 week period was comparable to one of our previous studies using the same rat strain and timeframe (Mischel and Mueller, 2011). Recording daily running distances (Figure 2a) and daily running times (Figure 2b) revealed that rats initially used the running wheel for a total of 45 min/day and ran just over 1 km/day; running then peaked at slightly over 3 km/day and 85 min/day of total running time. Eventually, running reached a steady-state, with rats using the wheel for an average of 75 min/day, achieving an average total distance approaching 2.8 km/day.

Figure 2. Running wheel activity.

Average daily running distance (a) and average daily running duration (b) for each week over the time course of the study. Active rats increased the distance and time that they spent using the running wheel over the first four weeks, with distance and duration peaking at five weeks. From the 6^th week on, the rats continued to run for a fairly constant but slightly decreasing distance and duration until the end of the study at 10–12 weeks. N values for rat numbers at 9, 10 and 11 week time points were 27, 22 and 13, respectively, reflecting the removal of animals from the data set for perfusion and fresh tissue harvesting. km = kilometers; min = minutes.

3.2. Validation of antibodies and injection sites

Western blot analyses of antibodies used in this study (Figure 3) revealed single bands at the appropriate and respective molecular weights for the three GABA receptors subunit antibodies (GABA_Aα1, ~50kDa Figure 3a; GABA_Aα2 (rabbit and mouse) ~55kDa; Figure 3b & 3c, respectively), KCC2 (~140 kDa; Figure 3d) and GAPDH (~38kDa; Figure 3e). To provide additional validation for the rabbit GABA_Aα1 antibody, preadsorption controls were carried out and abolition of immunostaining was confirmed when the antibody was preincubated with the control peptide antigen (AGA-001, Alomone Labs, Jerusalem, Israel; data not shown).

Figure 3. Validating antibodies and defining loading conditions.

Antibodies for biochemical and immunohistochemical studies were validated by western blotting. Single bands at expected molecular weights for each antigen indicate antibody specificity. a) The anti-GABA_Aα1 antibody was detected below the 52 kDa molecular weight marker, consistent with the molecular weight of 50kDa reported for GABA_Aα1. b & c) The anti-GABA_Aα2 antibodies (rabbit and mouse) were both detected just above the 52 kDa molecular weight marker, consistent with the molecular weight of ~55kDa reported for GABA_Aα2. d) The anti-KCC2 antibody was detected below the 150 kDa molecular weight marker, consistent with the molecular weight of 140 kDa reported for KCC2. e) The anti-GAPDH antibody was detected near the 38kDa molecular weight marker, consistent with the molecular weight of ~38kDa reported for GAPDH. f) Band intensities differ when different amounts of protein homogenate are loaded and probed using the GAPDH antibody shown in (e); increasing amounts of protein produce linear increase in band density on films scanned following ECL detection. R squared values demonstrate a high correlation between the amount of protein loaded and band density measured using Image J from the RVLMs of one sedentary (solid line, filled circles) and one active rat (dashed line, open circles). Vertical dotted lines represent the 5 μg of protein homogenate that was selected for protein loading.

Having established GAPDH as our loading control, we also verified that our loading conditions and the amount of protein homogenate per lane produced a linearly-related signal intensity upon densitometric analysis. For protein homogenates from both physically active and sedentary groups, 0.1–10 μg of loaded protein displayed a linear relationship in band intensity using the GAPDH antibody (Figure 3f).

Using immunofluorescent (Figure 4) or immunoperoxidase staining (not shown), we confirmed that our spinal cord injections of CTB (Figure 4a) were targeted to the intermediolateral cell column in spinal cord segments T9–10 (Figure 4b, 4c and 4c’) by comparing the brightfield image of our spinal cord section (Figure 4c) with the location of CTB-immunoreactivity in a fluorescence image of the same section (Figure 4c’). We used a brightfield image of each section (Figure 4d) to define a counting area that contained all of the bulbospinal C1 and bulbospinal non-C1 neurons (Figure 4d’) in the RVLM of that section. Using this strategy, we counted both types of bulbospinal neurons on a section-by-section basis.

Figure 4. Retrograde labelling and sectioning of the ventrolateral medulla for immunofluorescence studies.

a) Diagram showing how spinal cord injections of the retrograde tracer, cholera toxin B subunit (CTB), at thoracic levels nine and ten (T9–T10) retrogradely label neurons in the RVLM and RVLM_RE. b) Placement of CTB injections were confirmed with the aid of a rat atlas (Paxinos and Watson, 2007) to identify the intermediolateral cell column (IML). c and c’) A brightfield micrograph of a transverse section (c) shows the anatomical structures in that section, allowing precise definition of the location of CTB-immunoreactivity in a fluorescence micrograph of the same section (c’). d) (left) Tracing of a rat brainstem section labeled with the aid of a rat atlas (Paxinos and Watson, 2007); (right) Brightfield image of the same brainstem section. The rectangular box delineates the area in which bulbospinal catecholamine (C1) and non-catecholamine (non-C1) neurons were counted. (d’) Montage of fluorescent images taken from the boxed area in the section shown in (d). Neurons retrogradely labeled from T9–10 with cholera toxin B (CTB; fluorescence represented as blue); tyrosine hydroxylase (TH; red) and GABA_Aα2 receptor subunits (GABA_Aα2; green). e) Diagram showing the procedure used for obtaining sections through RVLM and RVLM_RE for immunofluorescence studies. Each brainstem was cut at 30 μm into a 1:4 series of sections and sections for staining were selected on the basis of their location relative to FN0 (depicted by the solid vertical line). The RVLM and the RVLM_RE were each divided into five regions (RVLM: FN-120 to FN-600); RVLM_RE: FN0 to FN+480) based on their locations caudal and rostral to FN0, respectively. NAc = compact division of the nucleus ambiguus; Sp5 = spinotrigeminal nucleus; Py= pyramids.

3.3. Western blotting for GABA receptor subunits

Consistent with our initial validation and a previous report (Foley et al., 2003), gel electrophoresis indicated that the GABA_Aα1 receptor subunit had a molecular weight of ~50 kDa (Figure 5a) and the GABA_Aα2 subunit, a molecular weight of ~55 kDa (Figure 5c). Both subunits were present at each level of the RVLM (FN-480 and FN-240) and RVLM_RE (FN+240 and FN+480) (Figures 5a and 5c).

Figure 5. Western blotting of GABA_Aα1 and GABA_Aα2 receptor subunits in the ventrolateral medulla of sedentary and physically active rats.

a) Examples of GABA_Aα1 expression at different rostrocaudal levels of the RVLM plus RVLM_RE in one sedentary and one active rat. b) Group data for GABA_Aα1 from sedentary and physically active conditions (n=12 rats per group). There were significant main effects of rostrocaudal location (F_3,66 = 21.2; p<0.001) but no significant main effect of sedentary versus physically active rats (F_1,66 = 2.3; p=0.143). A significant interaction (F_3,66 = 4.5; p=0.006) justified post hoc testing, which revealed significantly lower expression of GABA_Aα1 at −480 in sedentary compared to active rats (p=0.002, post hoc Holm-Sidak). Sedentary rats also demonstrated more rostrocaudal differences compared to active rats (see # and bars, see Appendix 1 for p values). c) Examples of GABA_Aα2 expression at different rostrocaudal levels of the RVLM plus RVLM_RE in one sedentary and one active rat. d) Group data for GABA_Aα2 from sedentary and physically active conditions (n=12 rats per group). There were significant main effects of both rostrocaudal location (F_3,66= 3.5; p=0.019) and group (F_1,66= 18.0, p<0.001), demonstrating an overall reduction in GABA_Aα2 subunit in sedentary animals. A lack of a significant interaction between rostrocaudal location and group (F_3,66=0.565; p=0.64) precluded post hoc testing. However, simple main effects testing within rostrocaudal location revealed a significantly lower expression of the GABA_Aα2 subunit at +480 μm when compared to +240 μm (#, p=0.002; Holm-Sidak) independent of sedentary or active conditions (see Appendix 1 for all simple main effects comparisons).

The GABA_Aα1 receptor subunit data demonstrated an overall main effect of rostrocaudal location (p<0.001; main effect); whereas, the main effect of sedentary versus physically active conditions did not reach significance (p=0.143, n=12 per group) (Figure 5b). Since a significant interaction term occurred (p=0.006), we deferred to post hoc testing, which revealed significantly less GABA_Aα1 receptor subunit at FN-480 in sedentary animals (*, Figure 5b; p=0.002). In addition, more prominent rostrocaudal effects occurred in sedentary rats; that is, post hoc testing revealed that sedentary conditions resulted in significantly reduced GABA_Aα1 subunit expression in caudal (i.e. RVLM) compared to rostral subregions, (i.e. RVLM_RE; Figure 5b; see Appendix 1 for all post hoc multiple comparisons). In contrast, active conditions only resulted in significantly lower GABA_Aα1 receptor subunit in the most caudal subregion, FN-480, compared to the most rostral subregion, FN+480 (Figure 5b; p=0.006).

Expression of the GABA_Aα2 subunit in the RVLM and RVLM_RE was significantly lower in sedentary animals (Figure 5d; p<0.001, main effect of group; n=12 per group), and although the main effect of rostrocaudal location was significant (p=0.019), post hoc testing of simple main effects revealed only a significantly lower expression in the most rostral subregion, FN+480 compared to FN+240 (see Appendix 1 for all simple main effects comparisons). Also, without a significant interaction term between group and subregion (p=0.64), further post hoc analysis was not justified.

Western blotting for the chloride transporter revealed that KCC2 was also present at each level of the RVLM (FN-480 and FN-240) and RVLM_RE (FN+240 and FN+480) (Figure 6a). Sedentary conditions resulted in a significant reduction in KCC2 compared to active conditions (Figure 6b; n=9 per group; p=0.043, main effect of group). There was also a significant main effect of rostrocaudal location (p<0.001); but because there was no significant interaction term (p= 0.215), further post hoc testing within groups across individual levels was not justified. Post hoc testing within the significant rostrocaudal main effect, did reveal that KCC2 expression was significantly lower in the more caudal subregions (i.e. RVLM) compared to the more rostral subregions (i.e. RVLM_RE; see Appendix 1 for all simple main effects comparisons). Interestingly, the rostrocaudal expression pattern of KCC2 resembled that of the GABA_Aα1 subunit more than the GABA_Aα2 subunit (compare Figure 6b to Figures 5b and 5d).

Figure 6. Western blotting analysis of KCC2 in the ventrolateral medulla of sedentary and physically active rats.

a) Examples of KCC2 expression at different rostrocaudal levels of the RVLM and RVLM_RE in one sedentary and one active rat. b) Group data from sedentary and physically active conditions (n=9 rats per group). KCC2 was significantly lower in sedentary compared to active rats (*, F_1,48 = 4.86; p=0.043, main effect of group) and demonstrated a significant main effect of rostrocaudal location (F_3,48 = 27.0; p<0.001). Post hoc testing within significant main effects indicated that KCC2 expression was significantly reduced in the more caudal subregions, FN-480 and FN-240 (i.e. RVLM) compared to the more rostral subregions, FN+240 and FN+480 (i.e. RVLM_RE) independent of sedentary or active conditions (#, see Appendix 1 for all simple main effects comparisons). Only the most caudal regions, FN-480 and FN-240, expressed similar levels of KCC2 compared to each other (p=0.21). There was no significant interaction between main effects (p=0.215; F_3,48= 1.55).

3.4. GABA_Aα subunits in bulbospinal C1 and non-C1 neurons

We readily identified spinally-projecting, C1 and non-C1 neurons by the presence of immunofluorescent staining for retrogradely transported CTB (fluorescence represented as blue; Figures 7a, 8a and 9a) and the presence or absence, respectively, of red fluorescent staining for TH (Figures 7b, 8b and 9b). Because green immunofluorescence indicating immunoreactivity for the GABA_Aα1 (Figure 7c and 8c) or GABA_Aα2 (Figure 9c) subunit was also clearly apparent across the RVLM and RVLM_RE, we were able to quantify the colocalization of the GABA_Aα1 subunit or the GABA_Aα2 subunit in the bulbospinal C1 and non-C1 neurons on a section-by-section basis.

Figure 7. Immunoreactivity for the GABA_Aα1 receptor subunit in bulbospinal C1 neurons in sedentary and physically active rats.

(a) Neuron retrogradely labeled from T9–T10 with cholera toxin B (CTB) fluorescence represented as blue. (b) Labeling of same neuron with anti-tyrosine hydroxylase reveals red, TH-positive C1 neuron (arrow). (c) Labeling of same neuron expressing GABA_Aα1 receptor subunits is shown with green immunofluorescence. (d) Merged image showing immunofluorescence for each of the three antigens. Boundaries of neuronal cell bodies are indicated by dashed lines. (e) Counts of GABA_Aα1-immunoreactive bulbospinal neurons are shown at each of the 10 rostrocaudal levels encompassing the RVLM plus RVLM_RE following 10–12 weeks of sedentary (filled circles) vs. physically active (open circles) conditions. There was no significant difference in the absolute number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 (F_1,135 =1.3; p=0.27) between sedentary and physically active rats. However, the absolute number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 did show a main effect of rostrocaudal distribution (#, F_9,135 =18.9; p<0.001). Post hoc testing for simple main effects of rostrocaudal location suggested an increasing number of bulbospinal C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα1-immunoreactive C1 neurons from caudal to rostral in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). There was no significant interaction term between group and rostrocaudal location (F_{9, 135} = 0.418; p=0.92). (f) In sedentary rats, a significantly higher percentage of spinally-projecting C1 neurons were immunoreactive for GABA_Aα1 when compared to physically active rats (*, F_1,135 = 5.3; p=0.037) and there was a main effect of rostrocaudal location (F_9,135 =8.6; p<0.001). Post hoc testing for simple main effects of rostrocaudal location suggested a decreasing percentage of bulbospinal C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a fairly stable number of bulbospinal GABA_Aα1-immunoreactive C1 neurons from caudal to rostral in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). There was no significant interaction term between experimental groups and rostrocaudal location (F_1,135 =1.1; p=0.36).

Figure 8. Immunoreactivity for the GABA_Aα1 receptor subunit in bulbospinal non- C1 neurons in sedentary and physically active rats.

(a) Neuron retrogradely labeled from T9T10 with cholera toxin B (CTB) fluorescence represented as blue. (b) Lack of labeling in the same neuron with anti-tyrosine hydroxylase reveals TH-negative non-C1 neuron. (c) Same neuron lacking expression of GABA_Aα1 receptor subunits. (d) Merged image showing only immunofluorescence for CTB. Boundaries of neuronal cell bodies are indicated by dashed lines. (e) Counts of GABA_Aα1-immunoreactive bulbospinal neurons are shown at each of the 10 rostrocaudal levels encompassing the RVLM plus RVLM_RE following 10–12 weeks of sedentary (filled circles) vs. physically active (open circles) conditions. In sedentary rats, there were significantly fewer GABA_Aα1-immunoreactive bulbospinal non-C1 neurons than in physically active animals, (*, F_1,135 = 7.4; p=0.016) and there was a main effect of rostrocaudal location (#, F_9,135 = 20.9; p<0.001). Post hoc tests within the main effect of rostrocaudal location suggested an increasing number of bulbospinal non-C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα1-immunoreactive C1 neurons from caudal to rostral in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). There was no significant interaction between group and rostrocaudal location (F_9,135 = 1.12; p=0.35). (f) A similar trend for a lower percentage of non-C1 cells immunoreactive for the GABA_Aα1 receptor subunit occurred in sedentary rats but did not reach significance (F_1,135 = 4.1; p=0.06). Similar to the absolute number of non-C1 neurons, there was a significant main effect of rostrocaudal location (#, F_9,135 = 14.2; p<0.001). Post hoc testing for simple main effects of rostrocaudal location suggested an increasing percentage of bulbospinal non-C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but little change in the number of bulbospinal GABA_Aα1-immunoreactive non-C1 neurons in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). There was no significant interaction term between group and rostrocaudal location (F_9,135 = 1.2; p=0.272).

Figure 9. Immunoreactivity for the GABA_Aα2 receptor subunit in bulbospinal C1 and non- C1 neurons in sedentary and physically active rats.

(a) Immunofluorescent neurons retrogradely labeled from T9–T10 with cholera toxin B (CTB; fluorescence represented as blue). (b) Labeling of the same set of neurons with anti-tyrosine hydroxylase reveals TH-immunoreactive C1 neurons with red immunofluorescence (arrows). TH-negative bulbospinal non-C1 neurons do not show red labelling (arrowheads). (c) Green immunofluorescent labelling identifies the same set of neurons immunoreactive for the GABA_Aα2 receptor subunit. (d) Merged images showing immunofluorescence for each of the three antigens. Double arrowheads designate neurons that show only immunoreactivity for GABA_Aα2. (e) Counts of GABA_Aα2-immunoreactive bulbospinal neurons are shown at each of the 10 rostrocaudal levels encompassing the RVLM plus RVLM_RE following 10–12 weeks of sedentary (black circles) vs. physically active (white circles) conditions. GABA_Aα2 receptor subunit immunoreactivity in bulbospinal C1 neurons following 10–12 weeks of sedentary (filled circles) vs. physically active (open circles) conditions at different rostrocaudal levels of the RVLM plus RVLM_RE. Sedentary conditions did not produce significant differences in the number of bulbospinal C1 neurons that were positive for the GABA_Aα2 subunit compared to active conditions (F_1,162 = 0.006; p=0.94) but a main effect of rostrocaudal location was evident (F_9,162 = 26.3; p<0.001). Post hoc tests within the main effect of rostrocaudal location suggested a fairly consistent distribution of bulbospinal C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα2-immunoreactive C1 neurons from caudal to rostral in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). A lack of significant interaction between group and rostrocaudal location (F_9,162 = 1.7; p= 0.09) precluded post hoc testing for group differences at specific rostrocaudal levels and suggested that the main effect of rostrocaudal location was driven by a higher number of bulbospinal GABA_Aα2-immunoreactive C1 neurons in the RVLM compared to the RVLM_RE. (f) The percentage of bulbospinal C1 neurons that show immunoreactivity for GABA_Aα2 subunit was not different between sedentary and active groups (F_1,162 = 1.19; p=0.291; respectively) but the percentage of bulbospinal C1 neurons that show immunoreactivity for GABA_Aα2 subunit demonstrated rostrocaudal location (F_9,162 = 15.76; p<0.001). Post hoc testing for simple main effects of rostrocaudal location suggested a decreasing percentage of bulbospinal C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but little change in the percentage of bulbospinal GABA_Aα2-immunoreactive C1 neurons in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). There was no significant interaction term between group and rostrocaudal location (F_9,162 = 0.57, p= 0.82). g) Differences in the absolute numbers of GABA_Aα2 immunoreactive bulbospinal non-C1 neurons in sedentary compared to active rats did not reach significance (F_1,162 =2.1; p=0.16). There was a significant effect of rostrocaudal location (#, F_9,162 =18.7; p<0.001). Post hoc tests within the main effect of rostrocaudal location suggested an increasing number of bulbospinal non-C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα2-immunoreactive C1 neurons from caudal to rostral in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). There was no significant interaction between group and rostrocaudal location (F_9,162 =0.91; p=0.51). Post hoc pairwise, multiple comparison procedures at each rostrocaudal location demonstrated significant differences at several levels (see Table 4). h) Differences in the percentages of non-C1 cells immunoreactive for the GABA_Aα2 subunit in sedentary versus physically active rats did not reach significance (F_1,162 = 2.3; p=0.15) but there was a significant effect of rostrocaudal location (#, F_9,162 = 17.9; p<0.001). Post hoc testing for simple main effects of rostrocaudal location suggested an increasing percentage of bulbospinal non-C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but little change in the number of bulbospinal GABA_Aα2-immunoreactive non-C1 neurons in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). There was no significant interaction between group and rostrocaudal location (F_9,162 =0.6; p=0.79).

3.4.1. GABA_Aα1 subunit immunofluorescence in bulbospinal C1 and non-C1 neurons

GABA_Aα1 in bulbospinal C1 neurons.

Bulbospinal C1 neurons containing the GABA_Aα1 subunit were distributed broadly across the rostrocaudal extent of the portion of the ventrolateral medulla examined in this study (Figure 7e). Although the absolute number of bulbospinal C1 neurons that were positive for the GABA_Aα1 subunit appeared to be slightly greater in sedentary rats (n=9) compared to active rats (n=8), the difference between groups did not reach significance (p=0.27, main effect of group) using two-way mixed ANOVA. There was however, a significant main effect of rostrocaudal location (p<0.001). In fact, post hoc testing for simple main effects of rostrocaudal location indicated an increasing number of bulbospinal C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm), and a decreasing number of bulbospinal GABA_Aα1-immunoreactive C1 neurons from caudal to rostral in the sections designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). Although a lack of a significant interaction term between main effects of group and subregion (p=0.92) precluded post hoc testing of group differences at each rostrocaudal location, the finding that bulbospinal GABA_Aα1-immunoreactive C1 neurons were differentially distributed prompted us to directly assess the total number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 in the RVLM versus the RVLM_RE in sedentary and active rats (Table 4). Indeed, there were significantly fewer bulbospinal C1 neurons immunoreactive for GABA_Aα1 in the RVLM_RE compared to the RVLM (p=0.007, main effect of subregion). However, there was no difference in the total number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 between sedentary and active rats (p=0.269; main effect of group). The subregion by group interaction was not significant (p=0.92), which precluded post hoc testing.

Table 4.

Number of Bulbospinal Neurons with GABA_A Subunit-Immunoreactivity

	RVLM (FN-5 to FN-1)		RVLMRE (FN0 to FN+4)
Bulbospinal C1 = CTB+ / TH+	Active	Sedentary	Active	Sedentary
GABAAα1+ (n = 8/9)a,b	46±5	51±5	38±4#	42±3#
GABAAα2+ (n = 10/10)c,d	79±3	79±4	49±3#	50±3#
Bulbospinal non-C1 = CTB+ / TH−	Active	Sedentary	Active	Sedentary
GABAAα1+ (n = 8/9)e,f	29±4	19±3*	44±6#	38±4#;*
GABAAα2+ (n = 10/10)g,h	23±3	16±4	34±2#	31±4#

Data presented as mean ± SEM; 2-way mixed ANOVA for all comparisons (see p and f values below).,

C1/GABA_Aα1+

^aF_1,15 = 1.3; p=0.269, Active vs Sedentary;

^b#, F_1,15 = 9.9, p=0.007, RVLM vs RVLM_RE; Intxn: F_1,15 = 0.011, p=0.92

C1/GABA_Aα2+

^cF_1,15 = 0.023; p= 0.88; Active vs Sedentary

^d#, F_1,15 = 57.8, p<0.001, RVLM vs RVLM_RE; Intxn: F_1,15 = 0.15, p=0.70

non-C1/GABA_Aα1+

^e*, F_1,15 = 6.3, p=0.025; Active vs Sedentary;

^f#, F_1,15 = 13.8, p=0.002, RVLM vs RVLM_RE; Intxn: F_1,15 = 0.097, p=0.76

non-C1/GABA_Aα2+

^gF_1,15 = 2.3, p=0.15; Active vs Sedentary;

^h#, F_1,15 = 27.7, p<0.001, RVLM vs RVLM_RE, Intxn: F_1,15 = 1.45, p=0.24

Because of the possibility that the total number of bulbospinal neurons immunoreactive for GABA_A subunits could be different between sedentary and active animals, we also computed the number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 as a percentage of the total number of bulbospinal C1 neurons on a section-by-section basis (Figure 7f), and directly compared these percentages in the RVLM versus RVLM_RE by totaling the number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 in the RVLM and RVLM_RE separately and expressing them as a percentage of the total number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 (Table 5).

TABLE 5.

Percentages of Bulbospinal Neurons with GABA_A Subunit-Immunoreactivity

	RVLM (FN-5 to FN-1)		RVLMre (FN0 to FN+4)
Bulbospinal C1 = CTB+ / TH+	Active	Sedentary	Active	Sedentary
GABAAα1+ (n = 8/9)a,b	55±4%	66±3%*	45±3%#	49±3%#*
GABAAα2+ (n = 10/10)c,d	76±2%	80±3%	58±2%#	61±3%#
Bulbospinal non-C1 = CTB+ / TH−	Active	Sedentary	Active	Sedentary
GABAAα1+ (n = 8/9)e,f	29±3%	20±2%p=0.06	43±3%#	41±4%#;p=0.06
GABAAα2+ (n = 10/10)g,h	18±2%	12±3%	37±2%#	32±4%#

Data presented as mean ± SEM; 2-way mixed ANOVA for all comparisons (see p and f values below).

C1/GABA_Aα1+

^a*, F_1,15 = 5.3; p=0.037; Active vs Sedentary;

^b#, F_1,15 = 21.2; p<0.001, RVLM vs RVLM_RE; Intxn: F _1,15 = 1.54, p=0.23

C1/GABA_Aα2+

^cF_1,15 = 0.9; p=0.35; Active vs Sedentary;

^d#, F_1,15 = 81.9; p<0.001; RVLM vs RVLM_RE; Intxn: F_1,15 = 0.00003, p=0.99

non-C1/GABA_Aα1+

^eF_1,15 = 4.1; p=0.06; Active vs Sedentary;

^f#, F_1,15 = 36.7; p<0.001; RVLM vs RVLM_RE; Intxn: F_1,15 = 1.45, p=0.25

non-C1/GABA_Aα2+

^gF_1,15 = 2.3; p=0.146; Active vs Sedentary;

^h#, F_1,15 = 27.4; p<0.001; RVLM vs RVLM_RE; Intxn: F_1,15 = 0.24, p= p=0.63

Sedentary conditions (n=9) versus active conditions (n=8) resulted in a significantly greater percentage of bulbospinal C1 neurons that were positive for the GABA_Aα1 subunit compared to physically active conditions (Figure 7f; p=0.037 main effect of group). In addition, the percentage of bulbospinal C1 neurons that were positive for the GABA_Aα1 subunit varied as a function of rostrocaudal location (Figure 7f; p<0.001, main effect of section number). Post hoc testing for simple main effects of rostrocaudal location suggested a decreasing percentage of bulbospinal C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but a fairly stable number of bulbospinal GABA_Aα1-immunoreactive C1 neurons from caudal to rostral in sections designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). Although a lack of a significant interaction term between main effects of group and subregion (p=0.36) precluded post hoc testing of group differences at each rostrocaudal level, the finding that bulbospinal GABA_Aα1-immunoreactive C1 neurons were differentially distributed prompted us to further directly assess the percentage of bulbospinal C1 neurons immunoreactive for GABA_Aα1 in the RVLM versus the RVLM_RE in sedentary and active rats (Table 5).

There was a significantly lower percentage of bulbospinal C1 neurons immunoreactive for GABA_Aα1 in the RVLM_RE compared to the RVLM (Table 5; p<0.001, main effect); and a higher percentage of bulbospinal C1 neurons immunoreactive for GABA_Aα1 in sedentary compared to active rats (p=0.037, main effect of group). A lack of a significant interaction term between group and subregion precluded post hoc testing (p=0.232).

GABA_Aα1 in bulbospinal non-C1 neurons.

Bulbospinal non-C1 neurons containing the GABA_Aα1 subunit were also distributed across the rostrocaudal extent of the ventrolateral medulla (Figure 8e). Sedentary conditions (n=9) versus active conditions (n=8) resulted in significantly fewer bulbospinal non-C1 neurons expressing the GABA_Aα1 subunit (Figure 8e; p=0.016, main effect of group). There was also a significant main effect of rostrocaudal location (p<0.001 for main effect of section number) but the interaction term between group and rostrocaudal location was not significant (p=0.35). Post hoc tests within the main effect of rostrocaudal location indicated an increasing number of bulbospinal non-C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα1-immunoreactive non-C1 neurons from caudal to rostral in the sections designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons).

The total number of non-C1 neurons immunoreactive for GABA_Aα1 was higher in the RVLM_RE when compared to the RVLM (Table 4; p=0.002, main effect for subregion). In addition, sedentary conditions resulted in a smaller total number of bulbospinal non-C1 neurons immunoreactive for GABA_Aα1 (Table 4; p=0.025, main effect of group). A lack of significant interaction (p=0.76) precluded post hoc testing and suggested an overall increase in the number of non-C1 neurons immunoreactive for GABA_Aα1 in the RVLM_RE similar to the section-by-section analysis and fewer non-C1 neurons immunoreactive for GABA_Aα1 in sedentary rats across both the RVLM and the RVLM_RE.

Because we could calculate the number of bulbospinal C1 neurons immunoreactive for GABA_Aα1 as a percentage of the total number of bulbospinal neurons, we also examined the number of non-C1 neurons immunoreactive for GABA_Aα1 as a percentage of the total number of bulbospinal neurons on a section-by-section basis (Figure 8f) and also directly compared the resulting percentages in the RVLM and RVLM_RE (Table 5). A difference in the percentage of bulbospinal non-C1 neurons positive for GABA_Aα1 subunits between sedentary (n=9) and active (n=8) animals did not quite reach significance (p=0.06, main effect of group) but there was a significant main effect of rostrocaudal location (p<0.001). Post hoc testing for simple main effects of rostrocaudal location suggested an increasing percentage of bulbospinal non-C1 neurons with GABA_Aα1-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but little change in the percentage of bulbospinal GABA_Aα1-immunoreactive non-C1 neurons in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 2 for all simple main effects comparisons). Also, based on the lack of a significant interaction (p=0.272), the significant rostrocaudal effect appeared to be driven by lower percentages of bulbospinal GABA_Aα1-positive non-C1 neurons caudally and higher percentages rostrally in the RVLM/RVLM_RE. In fact, when compared directly, the RVLM_RE contained a higher percentage of bulbospinal GABA_Aα1-positive non-C1 neurons than the RVLM (Table 5; p<0.001; main effect of subregion) and sedentary animals (n=9) had a strong trend for a lower percentage of bulbospinal non-C1 neurons positive for the GABA_Aα1 subunits compared to active animals (n=8; p=0.06; main effect of group). A lack of significant interaction between subregion and group (p=0.25) precluded post hoc testing and suggested that the lower percentages of bulbospinal GABA_Aα1-immunoreactive non-C1 neurons was a general effect in sedentary animals and not specific to the RVLM_RE or RVLM even though the percentages were lower in the RVLM overall.

3.4.2. GABA_Aα2 subunit immunofluorescence on bulbospinal C1 and non-C1 neurons

We used the same strategy to study bulbospinal C1 and non-C1 neurons expressing the GABA_Aα2 receptor subunit (Figure 9). Throughout the ventrolateral medulla, there were neurons retrogradely labeled with CTB from T9–T10 (immunofluorescent staining for retrogradely transported CTB (fluorescence represented as blue; Figure 9a). These neurons either contained or lacked red fluorescent staining for TH (Figure 9b), denoting them as C1 or non-C1 neurons, respectively. Green fluorescent labeling for GABA_Aα2-immunoreactivity also occurred in neurons across the ventrolateral medulla (Figure 9c). As a result, we could identify single- (bulbospinal), double- (C1 and non-C1) and triple- (GABA_Aα2 subunit immunoreactive) labeled neurons. (Figure 9d).

GABA_Aα2 in bulbospinal C1 neurons.

Bulbospinal C1 neurons containing the GABA_Aα2 subunit were distributed broadly across the rostrocaudal extent of the portion of the ventrolateral medulla examined in this study (Figure 9e). Sedentary conditions did not produce significant differences in the number of bulbospinal C1 neurons that were positive for the GABA_Aα2 subunit compared to active conditions (n=10; p=0.94, main effect) but a main effect of rostrocaudal location was evident (p<0.001, main effect of section number). Post hoc tests within the main effect of rostrocaudal location suggested a fairly consistent distribution of bulbospinal C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα2-immunoreactive C1 neurons from caudal to rostral in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). A lack of significant interaction between group and rostrocaudal location (p=0.091) precluded post hoc testing for group differences at specific rostrocaudal levels and suggested that the main effect of rostrocaudal location was driven by a higher number of bulbospinal GABA_Aα2-immunoreactive C1 neurons in the RVLM compared to the RVLM_RE. This interpretation was supported by direct comparisons between the RVLM and RVLM_RE in sedentary versus physically active rats (Table 4). The total number of bulbospinal C1 neurons that expressed the GABA_Aα2 subunit was significantly lower in the RVLM_RE compared to the RVLM (Table 4; p<0.001; main effect of subregion) but was not significantly different between sedentary and active animals (p=0.88; main effect for group). There was also no significant interaction between subregion and group (p=0.70), supporting our contention that the total number of bulbospinal C1 neurons that expressed the GABA_Aα2 subunit was higher in the RVLM compared to the RVLM_RE and overall were not different between active and sedentary animals.

For completeness sake, we analyzed the percentage of bulbospinal C1 neurons that were immunoreactive for the GABA_Aα2 subunit as a function of the total number of bulbospinal C1 neurons and found a significant main effect of rostrocaudal location (Figure 9f; p<0.001, main effect of section number). However, sedentary conditions (n=10) did not produce significant differences in the percentage of triple-labeled neurons compared to active conditions (n=10; p=0.35, main effect of group). Post hoc testing for simple main effects of rostrocaudal location suggested a decreasing percentage of bulbospinal C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but little change in the percentage of bulbospinal GABA_Aα2-immunoreactive C1 neurons in the region designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). Also, since there was no significant interaction between group and subregion (p>0.05), Figure 9f appears to indicate that the percentages of bulbospinal C1 neurons that were positive for the GABA_Aα2 subunit were higher caudally compared to rostrally in the RVLM/RVLM_RE. To support this contention, we directly compared the percentages of bulbospinal GABA_Aα2-immunoreactive C1 neurons in the RVLM and RVLM_RE in sedentary versus active groups (Table 5). The percentages of these neurons were very similar overall between sedentary and active rats (p=0.35; main effect of group) but were much lower in the RVLM_RE compared to the RVLM (p<0.001; Table 5) in both groups. The lack of interaction between group and subregion (p=0.99) reinforced our suggestion that the number of C1 neurons expressing the GABA_Aα2 subunit was higher in the RVLM but was unaffected by sedentary versus physically active conditions.

GABA_Aα2 in bulbospinal non-C1 neurons.

Bulbospinal non-C1 neurons containing the GABA_Aα2 subunit were also distributed broadly across the rostrocaudal extent of the region of the ventrolateral medulla examined in this study. Sedentary conditions (n=10) compared to active conditions (n=10) did not alter the number of bulbospinal non-C1 neurons expressing the GABA_Aα2 subunit (p=0.16, main effect of group) but the population of bulbospinal GABA_Aα2-immunoreactive non-C1 neurons was distributed in a rostrocaudal-dependent manner (Figure 9g; p<0.001, main effect of section number). Post hoc tests within the main effect of rostrocaudal location suggested an increasing number of bulbospinal non-C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) and a decreasing number of bulbospinal GABA_Aα2-immunoreactive C1 neurons from caudal to rostral in the sections designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). A lack of significant interaction between group and subregion precluded post hoc testing between groups at specific rostrocaudal levels (p=0.51). Nevertheless, examination of the rostrocaudal location suggested a difference in the number of bulbospinal GABA_Aα2-positive non-C1 neurons in rostral compared to caudal sections through the RVLM/RVLM_RE. This interpretation was supported when direct comparisons revealed a significantly greater number of non-C1 neurons expressing the GABA_Aα2 subunit in the RVLM_RE compared to the RVLM (Table 4; p<0.001; main effect of subregion). Similar to the section-by-section analysis, sedentary conditions did not alter the total number of bulbospinal non-C1 neurons positive for the GABA_Aα2 subunit (p=0.15). Without a significant interaction between group and subregion (p=0.24), the analysis was consistent with a greater number of bulbospinal non-C1 neurons expressing the GABA_Aα2 subunit in the RVLM_RE, independent of whether rats were active or sedentary.

The percentage of bulbospinal non-C1 neurons that were immunoreactive for the GABA_Aα2 subunit also varied rostrocaudally Figure 9h; p<0.001, main effect of section number). However, sedentary conditions (n=10) did not produce a significant effect compared to active conditions (n=10) (Figure 9h; p=0.15, main effect of group) and there was no significant interaction between rostrocaudal location and group (p=0.79, interaction). Post hoc testing for simple main effects of rostrocaudal location suggested an increasing percentage of bulbospinal non-C1 neurons with GABA_Aα2-immunoreactivity from caudal to rostral in the sections defined as RVLM (FN-600 to FN-120 μm) but little change in the percentage of bulbospinal GABA_Aα2-immunoreactive non-C1 neurons in the sections designated RVLM_RE (FN0 to FN+480 μm) (see Appendix 3 for all simple main effects comparisons). Direct comparisons between the RVLM and RVLM_RE in sedentary and physically active animals revealed significantly higher percentage of bulbospinal non-C1 neurons immunoreactive for the GABA_Aα2 subunit in the RVLM_RE compared to the RVLM (Table 5; p<0.001; main effect of subregion). There was no effect of sedentary versus active conditions (p=0.146; main effect of group) nor was there a significant interaction between subregion and group (p=0.63), suggesting a greater percentage of non-C1 neurons expressing the GABA_Aα2 subunit in the RVLM_RE, independent of whether rats were active or sedentary.

3.4.3. Total numbers and percentages of bulbospinal C1 or non-C1 neurons independent of GABA receptor subunit staining.

Having separately counted the total numbers of bulbospinal C1 and non-C1 neurons in two different sets of sections (one for each GABA receptor subunit) from eight of the same sedentary and eight of the same physically active rats, we combined these counts to give us the total number of bulbospinal C1 and non-C1 neurons in one half of the right RVLM (Figure 10). The number of bulbospinal C1 neurons was not significantly different between sedentary and physically active rats (n=8 per group; p=0.475; Figure 10a), although in the RVLM, there were significantly more bulbospinal C1 neurons compared to the RVLM_RE (p<0.001 for main effect of subregion; Figure 10a). A lack of a significant interaction term (p=0.983) precluded further testing via post hoc analyses, which indicated that the finding of a higher number of C1 neurons in the RVLM compared to the RVLM_RE was independent of active versus sedentary conditions.

Figure 10. Comparisons of the number of bulbospinal C1 and non-C1 neurons in the RVLM and RVLM_RE of sedentary and active rats.

(a) Number of bulbospinal C1 neurons in sedentary (filled bars) versus physically active (open bars) rats in the RVLM and RVLM_RE. There were no significant differences between sedentary and active animals (F_1,14 =0.54; p=0.475) but there was a significant main effect of subregion (F_1,14 =52.4; p<0.001). The lack of a significant interaction term (F_1,14 =0.0004; p=0.983) precluded post hoc testing, indicating lower numbers of C1 neurons in the RVLM_RE compared to the RVLM (two way mixed ANOVA) independent of active and sedentary conditions. (b) Number of bulbospinal non-C1 neurons in sedentary versus physically active rats in the RVLM and RVLM_RE. Sedentary rats had fewer bulbospinal non-C1 neurons than active rats (main effect of group; F_1,14 =8.3; p=0.012) and there were more bulbospinal non-C1 neurons in the RVLM_RE compared to the RVLM in both groups (main effect of subregion; F_1,14 =18.2; p<0.001). A lack of a significant interaction term (F_1,14 =0.04; p=0.85), indicated more non-C1 neurons in the RVLM_RE compared to the RVLM, independent of sedentary versus physically active conditions. c) Total numbers of bulbospinal C1 and non-C1 neurons (combined) in sedentary versus physically active rats in the RVLM and RVLM_RE. Sedentary rats had significantly fewer bulbospinal neurons than active rats (main effect of group; F_1,14 =8.1; p=0.013) but there was no significant differences between RVLM and RVLMRE in the total number of bulbospinal neurons (main effect of subregion; F_1,14 =1.45; p=0.248;). A lack of a significant interaction term between group and subregion (F_1,14 =0.025; p=0.89), precluded post hoc testing.

The combined counts of bulbospinal non-C1 neurons revealed that there were significantly fewer bulbospinal non-C1 neurons in sedentary animals (p=0.012; main effect of group; Figure 10b) and that there were significantly more of these neurons in the RVLM_RE compared to the RVLM (p<0.001; main effect of subregion; Figure 10b). Because there was no significant interaction term (p=0.85), these data suggested that the finding of fewer non-C1 neurons in sedentary animals occurred in both the RVLM and RVLM_RE and that the greater proportion of bulbospinal non-C1 neurons occurring in the RVLM_RE compared to the RVLM persisted in sedentary versus physically active animals.

Combining counts of both bulbospinal C1 and non-C1 neurons from sets of tissue from the same animals demonstrated that sedentary rats had significantly fewer bulbospinal neurons (n=8 per group; Figure 10c; p=0.013; main effect of group), but that in the RVLM, there was no significant difference in the number of bulbospinal neurons compared to the RVLM_RE (p=0.248; main effect of subregion). Because there was no significant interaction term between group and subregion (p=0.89), post hoc was not justified and indicated that the lower number of bulbospinal neurons (i.e., C1 neurons and non-C1 neurons) in sedentary compared to active rats was driven exclusively by a significant decrease in the number of spinally-projecting non-C1 neurons (Figures 10a versus 10b).

Lastly, because the number of bulbospinal non-C1 neurons was lower after sedentary conditions, the percentage of these neurons was also significantly lower in sedentary compared to active animals (n=8 per group; 28±3 versus 35±2%, respectively; p<0.05; t-test). And predictably, because of the decrease in the percentage of spinally-projecting non-C1 neurons, the percentage of bulbospinal C1 neurons was higher in sedentary compared to active rats (n=8 per group; 72±3 versus 65±2%; p<0.05; t-test).

4. Discussion

This study compared the influence of sedentary conditions on the relative amounts of two GABA_A receptor subunits as well as the locations and distributions of the neurons that express them in the RVLM (traditionally, the 500 μm caudal to the caudal pole of the facial nucleus (FN0)) (Verberne et al., 1999; Schreihofer and Sved, 2011) and its rostral extension, the RVLM_RE, named here and characterized here and elsewhere (Cano et al., 2004) (Minson et al., 1990; Stornetta and Guyenet, 1999). There are four important findings from this study. (1) Both spinally-projecting C1 and spinally-projecting non-C1 neurons differentially express GABA_A receptor subunits. Specifically, (2) sedentary conditions result in a significant increase in the percentage of spinally-projecting C1 neurons that express the GABA_Aα1 receptor subunit but a decrease in the absolute number of spinally-projecting non-C1 neurons that express this same GABA_Aα1 subunit. Also specifically, (3) GABA_Aα1 and GABA_Aα2 subunit expression vary significantly across the ventrolateral medulla, in patterns that differ between subunits and between sedentary compared to active animals. Finally, (4) the expression of the chloride transporter, KCC2, also increases from caudally in the RVLM to rostrally in the RVLM_RE and also, like both GABA_A receptor subunits, is significantly reduced by sedentary versus physically active conditions. Taken together, our studies suggest that both spinally-projecting C1 and non-C1 neurons that express GABA_Aα1 subunits exhibit (in)activity-induced neuroplasticity. We intentionally use the term “(in)activity-induced neuroplasticity” because it is currently unknown whether a lack of regular exercise promotes changes in GABA receptor subunits, whether regular physical activity alters GABA receptor expression or if each condition contributes to the relative differences observed in our cross-sectionally-designed study. Our data also show that GABA receptor subunits and KCC2, both of which play important roles in chloride-mediated inhibition of neuronal activity, decrease following sedentary conditions but that their overall amounts progressively increase from the RVLM to the RVLM_RE in both sedentary and active rats. It is possible that these differences across the RVLM and RVLM_RE reflect subregional alterations in the composition of GABA_A receptors and their interactions with KCC2.

Not only does our data reinforce the importance of examining structural, functional and cellular neuroplasticity within subregions of the RVLM (Mischel et al., 2015), our new findings significantly increase our understanding of how inactivity may change neurotransmission through the RVLM. First, our results are evidence for novel mechanisms that may explain both enhanced sympathoexcitatory and enhanced sympathoinhibitory responses generated by excitation and inhibition of the RVLM in sedentary versus physically active rats (Mischel and Mueller, 2011; Subramanian and Mueller, 2016; Dombrowski and Mueller, 2017). Our observations also provide an anatomical basis for enhanced sympathoexcitatory responses during reflex changes in sympathetic outflow documented previously in sedentary animals (DiCarlo and Bishop, 1988; Negrao et al., 1993). Given the strong relationship between elevations in sympathetic outflow and cardiovascular disease (Fisher et al., 2009; Malpas, 2010), our current and previous studies as well as those from other laboratories increase support for inactivity-related neuroplasticity as an important contributor to the development of cardiovascular disease in sedentary individuals (Michelini and Stern, 2009; Mueller, 2010; Martins-Pinge, 2011; Mischel et al., 2015; Mueller et al., 2017).

4.1. GABA_A receptor subunit localization on bulbospinal C1 and non-C1 neurons

In the present study, GABA_A receptor subunits were localized on bulbospinal C1 and non-C1 neurons in rats that were either sedentary or physically active. To our knowledge, this is the first study to report the presence of GABA receptor subunits on bulbospinal RVLM neurons, an unsurprising finding given the number of functional, biochemical and molecular publications on GABA neurotransmission in the RVLM over the past 35 years. For example, microinjections of GABA receptor agonists and antagonists have demonstrated the critical role of the RVLM in control of resting sympathetic outflow, as well as the prominent role of tonic inhibition on control of RVLM neurons by GABA_A receptors (Willette et al., 1983; Smith and Barron, 1990; Moffitt et al., 2002; Horiuchi et al., 2004; Schreihofer et al., 2005; Kvochina et al., 2007). Previous Western blotting and ELISA studies have also confirmed that tissue punches of the RVLM contain GABA_A receptor subunits, including the α1 and α2 subunits examined in the present study (Foley et al., 2003; Pawar et al., 2017; 2018). In addition, RT-PCR has identified mRNA for several GABA_A receptor subunits in RVLM tissue punches (Foley et al., 2003; Pawar et al., 2017; 2018); and, more recently, we detected mRNA for GABA receptor subunits in spinally-projecting RVLM neurons using laser capture microdissection and single cell qRT-PCR (Subramanian et al., 2014). The present study extends these findings to report that GABA_Aα1 and GABA_Aα2 receptor subunits are localized specifically on spinally-projecting RVLM neurons of both catecholamine (C1) and non-catecholamine (non-C1) subtypes. Furthermore, our western blotting data showing lower expression of both GABA_Aα1 receptor subunits in the RVLM and GABA_Aα2 receptor subunits in the RVLM and RVLM_RE in sedentary rats versus physically active rats are consistent with our gene expression studies of GABA_A receptors in individual bulbospinal RVLM neurons (Subramanian et al., 2014) and are another demonstration of activity-dependent neuroplasticity in the RVLM. A caveat for our studies, which should be stated explicitly, is that tissue punches from any brain region contain a mixed population of neuronal and non-neuronal cells. Thus, direct correlations between our western blotting and immunofluorescence results are necessarily imprecise.

Our immunofluorescence studies were purposely targeted at bulbospinal neurons projecting to the T9/10 area of the spinal cord where they innervate sympathetic preganglionic neurons that are in pathways regulating the splanchnic circulation. Focusing on this subset of bulbospinal neurons allowed us to provide anatomical correlates for our previous reports in which enhanced splanchnic sympathetic nerve responses were observed in sedentary versus physically active animals either after direct microinjections of glutamate or during decreases in blood pressure (Mischel and Mueller, 2011; Subramanian and Mueller, 2016). A challenge in this approach is that we cannot comment on RVLM neurons that exclusively innervate other regions of the spinal cord because our injections did not label these neurons. Another caveat is that, although GABA_Aα subunits are believed to be integral to a fully functioning GABA receptor (Olsen and Sieghart, 2009; Sieghart and Sperk, 2002), subunit expression is merely a surrogate measure for intact, functioning GABA receptors. In any case, as described in detail below, differential alterations in GABA receptors on bulbospinal C1 and non-C1 neurons are likely to be physiologically significant because both subsets of neurons plays important and different roles in control of sympathetic outflow (Dampney et al., 2003a; Guyenet, 2006; Schreihofer and Sved, 2011).

Interestingly, the two GABA receptor subunits examined in the present study varied in their expression levels and in their distribution rostrocaudally across the RVLM/RVLM_RE, including their expression on spinally-projecting C1 versus non-C1 neurons. While differences in the expression of one subunit could suggest a differential distribution of GABA_A receptors in the RVLM and RVLM_RE and their location on C1 and non-C1 neurons, differences in the expression of two different subunits could be indicative of alterations in GABA receptor subunit composition. GABA_A receptors have a pentameric structure and form voltage gated chloride ion channels, with the most prevalent subunit composition in the adult brain reported as α₁, β₂ and γ₂ subunits (Olsen and Sieghart, 2009). In contrast, α₂ subunits are confined to particular brain regions, where they can be expressed in patterns similar to α₁ subunits (Sieghart and Sperk, 2002; Olsen and Sieghart, 2009). Different GABA_Aα subunit compositions in sedentary versus active rats could change the affinity of the receptor for GABA and its activation and deactivation rates, thereby altering the properties of the chloride ion channel and leading to GABA_A receptors that function differently in sedentary versus active rats. Interestingly, GABA_A receptors with a specific subunit composition have recently been shown to affect the extent of dendritic branching and the number of spines on hippocampal neurons (Parato et al., 2018), suggesting that the types of GABA_A receptors on neurons can influence their morphology. We have previously shown that the dendritic trees of spinally-projecting C1 neurons differ between sedentary and active rats and it is tempting to speculate that the GABA_A receptor subtypes present on these neurons under sedentary versus physically active conditions could contribute to these dendritic differences.

In the present study, the differential expression of GABA_A subunits may also be related to subregional specificity in the control of sympathetic outflow by the RVLM, which was first reported over thirty years ago (Dampney and McAllen, 1988; McAllen and Dampney, 1990; McAllen and May, 1994). More recently, we have reported both structural and functional neuroplasticity in subregions of the rat RVLM, including in the region rostral to FN0, designated here as the RVLM_RE (Mueller et al., 2011; Mischel et al., 2014; Subramanian and Mueller, 2016; Huereca et al., 2018). Studies, including our own, on subregional differences have primarily focused on glutamatergic mechanisms of excitation in the RVLM (Mueller et al., 2011; Mischel et al., 2014; Subramanian and Mueller, 2016) and increases in dendritic branching. Perhaps prematurely, we attributed enhanced sympathoexcitation in sedentary animals to an increase in excitatory input (Mischel et al., 2014). Taking the results of the current study into account, it will now be important to consider the functional implications of differences in GABA receptors across the rostrocaudal extent of the RVLM, particularly since inactivity-related enhancements in GABAergic transmission have only recently been proposed and demonstrated (Mueller and Mischel, 2012; Dombrowski and Mueller, 2017).

4.2. Sedentary conditions differentially affect GABA-R subunit expression on C1 and non-C1 neurons

Sedentary conditions resulted in an increase in the percentage of spinally-projecting C1 neurons that express the GABA_Aα1 receptor subunit but a relative decrease in the absolute number of spinally-projecting non-C1 neurons that express this same GABA_Aα1 subunit (Figures 7f and 8e, respectively). Several previous studies have indicated that there are functionally relevant changes in GABAergic transmission in the RVLM under a variety of physiological and pathophysiological conditions. For example, inhibitory responses to GABA microinjections in the RVLM are enhanced in rats fed a high salt diet (Adams et al., 2007) and inhibitory responses to unilateral microinjections of GABA into the RVLM are enhanced in sedentary versus physically active rats (Dombrowski and Mueller, 2017). Interestingly, in the later study, sedentary animals showed enhanced responses to GABA only after inhibition of the contralateral RVLM, presumably because under intact conditions the contralateral side was able to compensate for decreases in sympathetic nerve activity. These latter findings substantiate why withdrawal of tonic, GABAergic inhibition of the RVLM is considered the primary mechanism that produces sympathoexcitation during baroreceptor unloading (Guyenet, 2006; Mueller, 2007; Schreihofer and Sved, 2011). It is well established that sedentary conditions result in enhanced baroreflex-mediated sympathoexcitation (DiCarlo and Bishop, 1988; Negrao et al., 1993; Mischel and Mueller, 2011), and it is therefore consistent that more bulbospinal C1 neurons expressing GABA_A receptors occur in the RVLM of sedentary animals. In other words, a higher percentage of spinally-projecting C1 neurons that express GABA_A receptors could provide an anatomical basis for enhanced sympathoexcitatory responses that occur during withdrawal of GABAergic input. Indeed, functional neuroanatomical studies using Fos, a marker of neuronal activation, have shown that hypotensive stimuli activate C1 neurons, including spinally projecting C1 neurons (Sved et al., 1994; Chan and Sawchenko, 1995; Graham et al., 1995; Minson et al., 1997; Dampney et al., 2003b). Furthermore, the number of Fos-positive neurons in sedentary animals is greater following other sympathoexcitatory stimuli, including acute exercise (Ichiyama et al., 2002) and restraint stress (Greenwood et al., 2003), with the latter study reporting a greater number of Fos-positive C1 neurons in sedentary animals. Whether the same hypotensive stimulus that produces enhanced sympathoexcitation in conscious sedentary animals (DiCarlo and Bishop, 1988; Negrao et al., 1993; Mischel and Mueller, 2011) also activates more C1 (and/or non-C1) neurons in the RVLM and RVLM_RE in sedentary versus physically active rats has yet to be tested.

An increased percentage of bulbospinal neurons expressing the GABA_Aα1 receptor subunit in the RVLM of sedentary rats could also explain the greater GABAergic modulation of glutamate-mediated sympathoexcitation in sedentary compared to physically active animals (Mueller and Mischel, 2012). For example, our previous study reported augmented sympathoexcitatory responses to glutamate microinjections in the RVLM in the presence of GABA_A receptor blockade in sedentary but not physically active rats (Mueller and Mischel, 2012), again suggestive of altered GABAergic signaling in the RVLM of sedentary animals. Because we found that fewer non-C1 neurons expressed GABA_Aα1 receptor subunits in the RVLM of sedentary rats in the present study, our hypotheses above presumes that C1 neurons mediate a greater proportion of sympathoexcitation during conditions of increased glutamatergic transmission. In fact, Guyenet and colleagues have hypothesized that C1 neurons play a role primarily in sympathoexcitatory states, where they serve to augment sympathoexcitation under a variety of conditions (Guyenet et al., 2013). Certainly, further studies are necessary to investigate these hypotheses more thoroughly.

In contrast to the increase in the percentage of C1 neurons expressing GABA_Aα1 subunits in sedentary animals, we observed a reduction in the absolute number of non-C1 neurons expressing GABA_Aα1 receptor subunits in sedentary rats. In addition, our western blot analysis showed a reduction in the GABA_Aα2 subunit content of the RVLM in sedentary compared to physically active rats. Taken together, these data could indicate less GABAergic inhibition on bulbospinal non-C1 neurons in sedentary animals. If that is the case, greater activation of bulbospinal non-C1 neurons (due to less GABA receptor-mediated inhibition) would be consistent with higher resting sympathetic outflow under sedentary conditions, as observed previously (Mischel and Mueller, 2011). Although it is controversial whether sedentary conditions universally augment sympathetic outflow in humans (Fagard, 2006; Carter and Ray, 2015), several animal studies report higher resting sympathetic outflow in sedentary versus physically active animals (Krieger et al., 2001; Mischel and Mueller, 2011). If, as proposed by Guyenet and colleagues (Guyenet et al., 2013), non-C1 neurons are primarily responsible for resting sympathetic outflow, it would be consistent that less tonic GABAergic inhibition of non-C1 neurons could drive an increase in resting sympathetic outflow in sedentary versus active animals (Krieger et al., 2001; Mischel and Mueller, 2011). However, it remains speculative whether reduced GABA receptor expression on non-C1 neurons is responsible for increases in resting sympathetic outflow in sedentary rats (Mischel and Mueller, 2011) and whether enhanced sympathoexcitation during withdrawal of GABAergic inhibition is generated primarily by bulbospinal C1 neurons, an increased percentage of which were shown to express GABA_Aα1 subunits in sedentary rats in the present study.

4.3. GABA_Aα1, but not GABA_Aα2 subunits vary differently across the rostrocaudal extent of the ventrolateral medulla as assessed by western blotting

Although the presence of other cell types within each tissue punch could contribute to differences in expression, our quantitative western blotting data strongly suggest that the expression of GABA_Aα1 and KCC2, but not GABA_Aα2, increases caudally from the RVLM to rostrally in the RVLM_RE. Consistent with this interpretation, our recent publications have highlighted similar variations in structure and function of RVLM neurons following sedentary conditions and reported a similar caudal to rostral variation (Mischel et al., 2015; Mueller et al., 2017). Previous findings that complement the reoccurring theme of region-specific, inactivity-related neuroplasticity in the ventrolateral medulla (i.e. RVLM and RVLM_RE) include: 1) increasing dendritic branching in bulbospinal C1 neurons from the RVLM to the RVLM_RE in sedentary but not physically active animals (Mischel et al., 2014); 2) enhanced splanchnic sympathetic nerve responses to direct glutamatergic activation of rostral versus caudal regions of the RVLM/RVLM_RE (Subramanian and Mueller, 2016); and 3) increasing levels of neuronal activity spanning from the RVLM to RVLM_RE of conscious, unexercised rats as estimated by manganese-enhanced magnetic resonance imaging (Huereca et al., 2018). As a whole, these findings indicate that following sedentary conditions there is a progressive, caudal-to-rostral increase in neuroplasticity that is functionally significant in the regulation of sympathetic outflow. Whether sedentary conditions alter the excitability of individual bulbospinal C1 or non-C1 neurons needs to be tested directly.

4.4. Changes in KCC2

To our knowledge, this is the first study to report differences in the expression of the chloride ion transporter KCC2 in the RVLM of sedentary versus physically active animals. We chose to examine KCC2 because of its involvement in maintaining the chloride ion gradient in neurons and KKC2’s potential interactions with the GABA_A receptor, a chloride ion channel (Sigel and Steinmann, 2012). In the present study, KCC2 in sedentary animals showed an expression pattern similar to both GABA_A receptor subunits, with a progressive caudal-to-rostral decrease from RVLM to RVLM_RE similar to the GABA_Aα1 subunit. We speculate that reduced expression of KCC2 could result in a smaller chloride ion gradient, reducing the inhibitory actions of GABA. If the reduction was substantial, the inhibitory actions of GABA could be reversed to excitatory via production a GABA-mediated outward current, as has been reported previously (Sigel and Steinmann, 2012). Given that sedentary conditions produce opposing effects on bulbospinal C1 and non-C1 neurons, follow-up studies will be required to determine how reduced KKC2 expression contributes to the alterations in sympathetic outflow produced by each subset of neurons and across the rostrocaudal extent of the RVLM/RVLM_RE.

In summary, our data and those from previous studies have identified important mechanisms within the brainstem by which a lack of physical activity may increase the incidence of cardiovascular disease and by which a sedentary lifestyle has become the leading cause of premature death (Blair, 2009). Further examination of alterations in GABA receptors and KCC2 before and after sedentary conditions will be important for understanding whether remaining inactive produces significant changes in GABA receptors over time. It is also possible that maintaining an active lifestyle offsets changes that would occur in the brain under sedentary conditions or produces mechanistically different changes (e.g., by altering KCC2) compared to a sedentary lifestyle. In any case, determining whether sedentary or active conditions affect GABA receptors and KCC2 differentially or similarly will be a necessary prerequisite for determining the appropriate therapeutic targets for reversing inactivity-related cardiovascular diseases.

Abbreviations:

NA

nucleus ambiguus

Sp5

spinal trigeminal tract

FN

facial nucleus

FN0

caudal pole of the facial nucleus

Appendix 1

Post Hoc Testing for GABA_Aα1 between Sedentary and Active animals and Post Hoc Simple Main Effects testing within Rostrocaudal Level for GABA_Aα2 expression (see text and Figure 6 for reference)

	GABAAα1				GABAAα2
Rostrocaudal position	Sedentary	Active		Rostrocaudal position	Simple Main Effects
Comparison	p value	p value		Comparison	p value
FN −480 vs. FN −240	0.008 #	0.489		FN −480 vs. FN −240	0.912
FN −480 vs. FN +240	<0.001 #	0.071		FN −480 vs. FN +240	0.356
FN −480 vs. FN +480	<0.001 #	0.006 #		FN −480 vs. FN +480	0.030
FN −240 vs. FN +240	0.009 #	0.258		FN −240 vs. FN +240	0.302
FN −240 vs. FN +480	<0.001 #	0.035		FN −240 vs. FN +480	0.039
FN+240 vs. FN +480	0.024 #	0.317		FN+240 vs. FN +480	0.002 #

^#, denotes significance after correcting for multiple comparison using Holm-Sidak method.

Post Hoc Testing for Simple Main Effects within Rostrocaudal Level for KCC2 expression (see text and Figure 5 for reference)

Rostrocaudal position	KCC2 expression (Simple Main Effects)
Comparison	p value
FN −480 vs. FN −240	0.209
FN −480 vs. FN +240	<0.001 #
FN −480 vs. FN +480	<0.001 #
FN −240 vs. FN +240	<0.001 #
FN −240 vs. FN +480	<0.001 #
FN+240 vs. FN +480	<0.003 #

^#, denotes significance after correcting for multiple comparison using Holm-Sidak method.

Appendix 2

Post Hoc Testing for Simple Main Effects within Rostrocaudal Level for the absolute numbers and percentages of GABA_Aα1-Immunoreactive (IR) C1 and non-C1 Neurons (see Figure 7 for reference)

Rostrocaudal position	C1 Neurons		non-C1 Neurons
	Number with IR	Percent with IR	Number with IR	Percent with IR
Comparison	P	P	P	P
FN −600 vs FN −480	0.0014 #	0.1750	0.9660	0.4640
FN −600 vs FN −360	0.0012 #	0.0878	<0.001 #	0.0812
FN −600 vs FN −240	<0.001 #	0.0173	<0.001 #	<0.001 #
FN −600 vs FN −120	<0.001 #	0.0012 #	<0.001 #	<0.001 #
FN −600 vs FN 0	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +120	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +240	0.0043	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +360	0.2870	0.0019	<0.001 #	<0.001 #
FN −600 vs FN +480	0.0608	0.0165	0.0707	<0.001 #
FN −480 vs FN −360	0.9540	0.0025	<0.001 #	0.0139
FN −480 vs FN −240	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN −120	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN 0	0.0036	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +120	0.6870	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +240	0.7300	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +360	0.0305	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +480	<0.001 #	<0.001 #	0.0643	<0.001 #
FN −360 vs FN −240	<0.001 #	0.4910	0.0022 #	0.0830
FN −360 vs FN −120	<0.001 #	0.1150	<0.001 #	0.0054
FN −360 vs FN 0	0.0043	0.0039	<0.001 #	<0.001 #
FN −360 vs FN +120	0.7300	<0.001 #	<0.001 #	<0.001 #
FN −360 vs FN +240	0.6870	0.0049	<0.001 #	<0.001 #
FN −360 vs FN +360	0.0264	0.1520	0.1430	0.0034
FN −360 vs FN +480	<0.001 #	0.4800	0.1070	0.0326
FN −240 vs FN −120	0.3070	0.3720	0.1160	0.2800
FN −240 vs FN 0	0.5360	0.0261	0.0011 #	0.0034
FN −240 vs. FN +120	0.0018 #	0.0072	0.0238	0.0012 #
FN −240 vs FN +240	<0.001 #	0.0317	0.1640	0.0042
FN −240 vs FN +360	<0.001 #	0.4550	0.1010	0.2180
FN −240 vs FN +480	<0.001 #	0.9860	<0.001 #	0.6800
FN −120 vs FN 0	0.1020	0.1790	0.0822	0.0599
FN −120 vs FN +120	<0.001 #	0.0690	0.4820	0.0285
FN −120 vs FN +240	<0.001 #	0.2050	0.8570	0.0703
FN −120 vs FN +360	<0.001 #	0.8840	0.0015 #	0.8780
FN −120 vs FN +480	<0.001 #	0.3810	<0.001 #	0.5030
FN 0 vs FN +120	0.0115	0.6310	0.2970	0.9420
FN 0 vs FN +240	0.0012 #	0.9370	0.0554	0.7520
FN 0 vs FN +360	<0.001 #	0.1360	<0.001 #	0.0835
FN 0 vs FN +480	<0.001 #	0.0273	<0.001 #	0.0113
FN +120 vs FN +240	0.4540	0.5770	<0.001 #	0.6970
FN +120 vs FN +360	0.0106	0.0498	0.3770	0.0413
FN +120 vs FN +480	<0.001 #	0.0075	<0.001 #	0.0046
FN +240 vs FN +360	0.0679	0.1580	<0.001 #	0.0971
FN +240 vs FN +480	<0.001 #	0.0331	0.0027 #	0.0138
FN +360 vs FN +480	0.0036	0.4650	0.0024 #	0.4110

^#, denotes significance after correcting for multiple comparisons using Holm-Sidak method.

Appendix 3

Post Hoc Testing for Simple Main Effects within Rostrocaudal Level for the absolute numbers and percentages of GABA_Aα2-Immunoreactive (IR) C1 and non-C1 Neurons (see Figure 8 for reference)

Rostrocaudal position	C1 Neurons		non-C1 Neurons
	Number with IR	Percent with IR	Number with IR	Percent with IR
Comparison	P	P	P	P
FN −600 vs FN −480	0.434	0.152	0.034	0.132
FN −600 vs FN −360	0.038	0.025	<0.001 #	0.014
FN −600 vs FN −240	0.024	0.007	<0.001 #	0.004
FN −600 vs FN −120	0.197	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN 0	0.838	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +120	0.241	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +240	0.009	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +360	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −600 vs FN +480	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN −360	0.191	0.410	0.061	0.330
FN −480 vs FN −240	0.135	0.190	0.010	0.169
FN −480 vs FN −120	0.611	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN 0	0.563	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +120	0.052	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +240	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −480 vs FN +360	<0.001 #	<0.001 #	0.011	<0.001 #
FN −480 vs FN +480	<0.001 #	<0.001 #	0.184	<0.001 #
FN −360 vs FN −240	0.851	0.625	0.474	0.685
FN −360 vs FN −120	0.423	<0.001 #	<0.001 #	<0.001 #
FN −360 vs FN 0	0.060	<0.001 #	<0.001 #	<0.001 #
FN −360 vs FN +120	0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −360 vs FN +240	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −360 vs FN +360	<0.001 #	<0.001 #	0.497	<0.001 #
FN −360 vs FN +480	<0.001 #	0.004 #	0.583	<0.001 #
FN −240 vs FN −120	0.323	0.001 #	0.002 #	<0.001 #
FN −240 vs FN 0	0.039	<0.001 #	<0.001 #	<0.001 #
FN −240 vs. FN +120	<0.001 #	<0.001 #	<0.001 #	<0.001 #
FN −240 vs FN +240	<0.001 #	<0.001 #	0.001 #	<0.001 #
FN −240 vs FN +360	<0.001 #	<0.001 #	0.971	<0.001 #
FN −240 vs FN +480	<0.001 #	0.015	0.207	0.003
FN −120 vs FN 0	0.278	0.096	0.133	0.111
FN −120 vs FN +120	0.0146	0.179	0.489	0.139
FN −120 vs FN +240	<0.001 #	0.114	0.880	0.026
FN −120 vs FN +360	<0.001 #	0.946	0.002 #	0.877
FN −120 vs FN +480	<0.001 #	0.427	<0.001 #	0.730
FN 0 vs FN +120	0.169	0.746	0.415	0.907
FN 0 vs FN +240	0.005 #	0.933	0.176	0.519
FN 0 vs FN +360	<0.001 #	0.110	<0.001 #	0.150
FN 0 vs FN +480	<0.001 #	0.015	<0.001 #	0.053
FN +120 vs FN +240	0.148	0.811	0.588	0.446
FN +120 vs FN +360	<0.001 #	0.201	<0.001 #	0.185
FN +120 vs FN +480	<0.001 #	0.033	<0.001 #	0.069
FN +240 vs FN +360	<0.001 #	0.130	0.001 #	0.038
FN +240 vs FN +480	<0.001 #	0.018	<0.001 #	0.010
FN +360 vs FN +480	0.023	0.389	0.220	0.617

^#, denotes significance after correcting for multiple comparisons using Holm-Sidak method.