Abstract
The mesencephalic locomotor region (MLR) plays a crucial role in locomotor control. In vertebrates, stimulation of the MLR at increasing intensities elicits locomotion of growing speed. This effect has been presumed to result from higher brain inputs activating the MLR like a dimmer switch. Here, we show in lampreys (Petromyzon marinus) of either sex that incremental stimulation of a region homologous to the mammalian substantia nigra pars compacta (SNc) evokes increasing activation of MLR cells with a graded increase in the frequency of locomotor movements. Neurons co-storing glutamate and dopamine were found to project from the primal SNc to the MLR. Blockade of glutamatergic transmission largely diminished MLR cell responses and locomotion. Local blockade of D1 receptors in the MLR decreased locomotor frequency, but did not disrupt the SNc-evoked graded control of locomotion. Our findings revealed the presence of a glutamatergic input to the MLR originating from the primal SNc that evokes graded locomotor movements.
SIGNIFICANCE STATEMENT The mesencephalic locomotor region (MLR) plays a crucial role in the control of locomotion. It projects downward to reticulospinal neurons that in turn activate the spinal locomotor networks. Increasing the intensity of MLR stimulation produces a growing activation of reticulospinal cells and a progressive increase in the speed of locomotor movements. Since the discovery of the MLR some 50 years ago, it has been presumed that higher brain regions activate the MLR in a graded fashion, but this has not been confirmed yet. Here, using a combination of techniques from cell to behavior, we provide evidence of a new glutamatergic pathway activating the MLR in a graded fashion, and consequently evoking a progressive increase in locomotor output.
Introduction
In the brainstem, the mesencephalic locomotor region (MLR) plays a crucial role in locomotor control. First discovered in cats by a Russian team (Shik et al., 1966), the MLR was found in all vertebrates tested afterward (lamprey: Sirota et al., 2000; salamander: Cabelguen et al., 2003; stingray: Bernau et al., 1991; bird: Sholomenko et al., 1991; rat: Garcia-Rill et al., 1987; mouse: Lee et al., 2014; Roseberry et al., 2016; rabbit: Musienko et al., 2008; guinea-pig: Marlinsky and Voitenko, 1991; monkey: Eidelberg et al., 1981; Karachi et al., 2010; Goetz et al., 2016). The MLR projects downward to reticulospinal neurons, which activate the spinal locomotor networks (cat: Orlovskiı̆, 1970; Steeves and Jordan, 1980; Garcia-Rill and Skinner, 1987a,b; Noga et al., 1991; Musienko et al., 2012; rat: Bachmann et al., 2013; bird: Sholomenko et al., 1991; lamprey: Buchanan and Grillner, 1987; Brocard et al., 2010; mouse: Bretzner and Brownstone, 2013; salamander: Ryczko et al., 2016a). One salient feature of the MLR lies in its ability to finely control locomotor output. Increasing MLR stimulation intensity produces a growing activation of reticulospinal cells and a progressive increase in the speed of locomotor movements.
Since the discovery of the MLR (Shik et al., 1966), it was presumed that the MLR is activated incrementally (i.e., like a dimmer switch) by higher brain regions, but the source of the rheostat-like inputs remained unknown. The MLR receives projections from the cortex, basal ganglia, periaqueductal gray, lateral hypothalamus (for review, see Ryczko and Dubuc, 2013), and from dopamine neurons of the substantia nigra pars compacta (SNc; Ryczko et al., 2013, 2016b). The physiological role of these different inputs to the MLR is not fully understood. The tonic inhibition sent by the output stations of the basal ganglia (lamprey: Stephenson-Jones et al., 2011, 2012; Ménard et al., 2007; cat: Takakusaki et al., 2003; mouse: Kravitz et al., 2010; Roseberry et al., 2016) is considered to be involved in action selection and therefore locomotion initiation or suppression, rather than fine control of locomotion (Albin et al., 1989; Redgrave et al., 1999; Grillner et al., 2013; Grillner and Robertson, 2016). The lamprey homolog of the mammalian motor cortex (i.e., pallium) was shown to elicit reticulospinal responses and locomotion (Ocaña et al., 2015), but its contribution to the graded control of MLR activity and locomotion is unknown.
So far, only one source of input to the MLR called the posterior tuberculum (PT) was characterized physiologically. This diencephalic region, which is considered homologous to the mammalian SNc (Pombal et al., 1997; Stephenson-Jones et al., 2011; Ryczko et al., 2013, 2016b), provides a strong excitatory input to MLR cells and robustly evokes locomotion (Derjean et al., 2010; Gariépy et al., 2012a; Ryczko et al., 2013). The PT sends a descending dopaminergic projection to the MLR (Ryczko et al., 2013; Pérez-Fernández et al., 2014) where it releases dopamine that increases locomotor output through a D1 receptor-dependent mechanism (Ryczko et al., 2013). Interestingly, the PT also contains glutamatergic neurons (Villar-Cerviño et al., 2011, 2013), and data suggest that their descending projections could control MLR activity. The PT sends non-dopaminergic (thus potentially glutamatergic) projections to the MLR (Ryczko et al., 2013). PT stimulation elicits fast synaptic responses in MLR cells (Gariépy et al., 2012a; Ryczko et al., 2013). Blockade of MLR glutamatergic receptors disrupts reticulospinal responses elicited by stimulation of the olfactory bulbs, which project to the PT (Derjean et al., 2010). Glutamate application in the MLR elicits locomotion (Brocard et al., 2010; Gariépy et al., 2012a). However, there is no direct evidence for descending glutamatergic inputs from the PT to the MLR, and it is unknown whether such input can progressively increase MLR activation and the speed of the locomotor movements. Here, using neural tracing, immunofluorescence, calcium imaging, patch-clamp recordings, and intracellular recordings during locomotion in lampreys, we show that a descending glutamatergic pathway originating from the primal SNc controls MLR activity and locomotor movements like a dimmer switch.
Materials and Methods
Ethics statement.
All procedures conformed to the guidelines of the Canadian Council on Animal Care and were approved by the animal care and use committees of the Université de Montréal (Quebec, Canada) and Université du Québec à Montréal (Quebec, Canada). A total of 47 sea lampreys (Petromyzon marinus) were used, with n = 36 larvae for physiology experiments and n = 4 transformed and n = 7 adults for anatomy experiments. Sex of the individuals used was not taken into account in the present study. Care was taken to minimize the number of animals used and their suffering.
Semi-intact and isolated brain preparations.
Larval sea lampreys were anesthetized with tricaine methanesulphonate (MS 222, 200 mg/L; Sigma-Aldrich) dissolved in a Ringer's solution (in mm: 130 NaCl, 2.1 KCl, 2.6 CaCl2, 1.8 MgCl2, 4.0 HEPES, 4.0 dextrose, and 1.0 NaHCO3 at pH 7.4). The animals were anesthetized for 8–10 min and then transferred into oxygenated cold Ringer's solution. To expose the brain, the skin and muscles were removed from the rostral part of the animal. The dorsal cranium was opened to obtain ad libitum access to the PT and to reticulospinal neurons in the middle rhombencephalic reticular nucleus (MRRN). Brain tissue rostral to the PT was removed by a transverse section between the diencephalon and the telencephalon. The rostral spinal segments were exposed and the caudal body parts were left intact. The brain was pinned down dorsal side up in the recording chamber, and the body was free to move in a video monitored chamber. To provide access to the PT, a dorsal midsagittal transection was performed at the level of the diencephalon. Recovery time lasted 1 h before the recording experiments began. For isolated brain preparations, the same dissection procedure was used, but the body was removed.
Electrophysiology and stimulation.
Intracellular recordings were performed with sharp glass microelectrodes (80–110 MΩ) filled with potassium acetate (4 m). The signals were amplified with an Axoclamp 2A (Molecular Devices). Only cells with a membrane potential <−60 mV and held stable for 15 min after impalement were included in the study.
Extracellular recordings of reticulospinal neurons were performed using glass micropipettes (diameter 5 μm) filled with Ringer's solution and the recordings were amplified with a model 1800 amplifier (bandwidth 100–500 Hz; A-M Systems). Signals were acquired (sampling rate of 5–10 kHz) through a Digidata 1200 series interface coupled with Clampex 9.0 (Brocard et al., 2010; Ryczko et al., 2013).
Targeted patch-clamp recordings of MLR cells were done using the protocol previously developed in the laboratory (Gariépy et al., 2012a; Ryczko et al., 2013). Briefly, MLR cells were first retrogradely labeled in vivo from an injection of Texas Red dextran amines (MW 3000 Da; Invitrogen) in the MRRN. The following day, the animal was killed and the brain was isolated and placed in a cold Ringer's solution (1–3 °C). To provide access to MLR cells, the dorsal part of the brain was removed with a vibratome. The giant reticulospinal cell I1 (Rovainen, 1967) was used as a landmark to locate the MLR (Brocard et al., 2010; Ryczko et al., 2013). Retrogradely labeled MLR cells were visualized under a microscope (Nikon Instruments or Olympus) equipped for fluorescence and targeted for whole-cell patch-clamp. Patch pipette (4–6 MΩ) were filled with a solution containing the following (in mm): 102.5 cesium methane sulfonate, 1 NaCl, 1 MgCl2, 5 EGTA, 5 HEPES, 0.3 ATP, and 0.1 GTP. The pH was adjusted to 7.2 with CsOH, and the osmolarity to 240 mOsm with H2O. The cellular electrophysiological signals were recorded with a model 2400 amplifier (A-M Systems). To measure the drug effects on the amplitude of synaptic responses, 20 excitatory postsynaptic potentials (EPSPs) were recorded for each cell and for each drug condition.
Glass-coated tungsten microelectrodes (0.7–3.1 MΩ with 10–40 μm exposed tip) and a Grass S88 stimulator (Astro Med) coupled to a Grass PSIU6 photoelectric isolation unit for controlling stimulation intensity (Astro Med) were used for electrical stimulation. The stimulation site was chosen on the basis of previous anatomical and physiological studies (Derjean et al., 2010; Gariépy et al., 2012a; Ryczko et al., 2013) and confirmed by subsequent histology (see Fig. 2B,C). The electrical stimulation consisted of square pulses (2 ms duration) applied with a frequency of 4–5 Hz for 10 s to elicit swimming. A pause of 3–5 min was made between two stimulations. Single pulses applied at a frequency of 0.1 Hz were used to evoke excitatory postsynaptic currents (EPSCs). High-frequency doublet pulses (20 Hz) were used to test for monosynaptic connectivity. The stimulation intensity ranged from 1 to 32 μA.
Ca2+ imaging.
MLR cells were retrogradely labeled by placing crystal of the Ca2+ indicator Ca2+ green-dextran amines (MW 3000, Invitrogen) at the level of the MRRN, immediately after a complete transverse transection of the brainstem at the level of the MRRN. The preparation was then transferred for 18–24 h in a chamber perfused with cooled (8–10°C), oxygenated Ringer's solution to allow the tracer to migrate to fill the MLR cell bodies. The next day, the brain tissue rostral to the PT was removed following a transverse section. As for the patch-clamp recording experiments described above, the dorsal part of the brain was cut away with a vibratome to provide access to MLR cells. The preparation was then pinned down to the bottom of a recording chamber perfused with cooled (8–10°C), oxygenated Ringer's solution (4 ml/min). Changes in fluorescence were recorded as previously (Brocard et al., 2010; Ryczko et al., 2016a,b) with a Nikon epifluorescent microscope coupled with a CCD video camera (Photometrics CoolSNAP HQ, Roper Scientific). To measure the changes in fluorescence, regions-of-interest were manually delineated around the MLR cell bodies labeled with the Ca2+ dye. Changes in fluorescence of MLR neurons to PT stimulation were acquired at a rate of 2 Hz using MetaFluor (Universal Imaging). The Ca2+ responses were expressed as the relative changes in fluorescence (ΔF/F). The baseline was defined as the averaged fluorescence before stimulation. Data analysis was performed using MetaFluor, Clampfit (Molecular Devices) and MATLAB (MathWorks). To measure drug effects on Ca2+ responses, 4–6 responses were recorded for each cell and for each drug condition.
Drug application.
Chemicals were purchased from Sigma-Aldrich and diluted to their final concentration in Ringer's solution. In some experiments, bath application of a Ringer's solution containing the AMPA/Kainate antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 25 μm) and the NMDA antagonist (2R)-amino-5-phosphonovaleric acid (AP5; 100 μm) was used. In some other experiments, a Ringer's solution containing CNQX (1 mm) and AP5 (0.5 mm; Gariépy et al., 2012a; Ryczko et al., 2016a) or the D1 receptor antagonist SCH 23390 (0.5 mm; Ryczko et al., 2013) was microinjected in the MLR. In some experiments, a Ringer's solution containing d,l-glutamate (2.5 mm) was also microinjected in the PT. The microinjection procedure was as previously described (Brocard and Dubuc, 2003; Le Ray et al., 2003; Derjean et al., 2010; Gariépy et al., 2012a, 2012b; Ryczko et al., 2013, 2016a,b). The microinjections were done with a glass micropipette (tip diameter of 10–20 μm) using pressure pulses (3–4 psi) of variable duration (10–80 ms) applied with a Picospritzer (General Valve). Fast green was added to the injected solution to monitor the extent of the injection site. The injected volumes were estimated by measuring the diameter of a droplet ejected in air from the tip of the pipette multiplied by the number of pressure pulses, and the resulting number of moles ejected was calculated for each drug (Le Ray et al., 2003; Ryczko et al., 2013, 2016a,b).
Kinematics.
Locomotor movements were monitored with a video camera (Sony HDR-XR200; 30 frames/s) positioned 1 m above the recording chamber. Data were analyzed offline using homemade software (Brocard et al., 2010; Gariépy et al., 2012a; Ryczko et al., 2013; Juvin et al., 2016). Briefly, tracking markers were equidistantly distributed along the body and monitored over time. Swimming was identified through the presence of mechanical waves traveling from head to tail (Fig. 1; Sirota et al., 2000; Ryczko et al., 2013). To quantify the frequency of swimming movements a single couple of markers located in the middle part of the body was used.
Anatomical tracing and immunofluorescence.
These experiments were performed on isolated brain preparations (see above). Biocytin (Sigma-Aldrich) was used for retrograde labeling of PT cells as previously described (Gariépy et al., 2012a,b, Ryczko et al., 2013, 2016a,b). First, a pulled glass micropipette was used to perform a lesion at the injection site in the MLR and crystals of biocytin were immediately placed in the lesion, allowing the dissolving tracer to be picked up by cut axons. After 10–15 min, the injection site was thoroughly rinsed and the brain was transferred to a chamber perfused with cold oxygenated Ringer's solution overnight to allow retrograde transport of the tracer. The injection site was chosen based on previous studies on the MLR (Derjean et al., 2010; Ryczko et al., 2013; for review see Ryczko and Dubuc, 2013) and verified by histology. The next day, the brain was transferred to a fixative solution according to the immunofluorescence procedure to follow. It is safe to presume that not all axons crossing the injection site were filled by the tracer. Consequently, the labeled cells in the PT constitute an underestimation of the actual number of cells projecting to the MLR in each preparation.
For tyrosine-hydroxylase (TH) immunofluorescence, the whole brain was immersed for 24 h in 4% (w/v) paraformaldehyde in PBS and then transferred to PB containing 20% (w/v) sucrose, both steps at 4°C. The next day, the brain was sectioned at 25 μm thickness with a cryostat. The sections collected on Color-Frost Plus slides (Fisher Scientific) were air-dried overnight at 37°C, after which they were rinsed three times 10 min in PBS and blocked in PBS containing 5% normal goat serum and 0.3% Triton X-100 for 60 min. The sections were then incubated overnight in the blocking solution containing the TH antibody. The next day, they were rinsed three times 10 min with PBS, incubated in the blocking solution containing the secondary antibody for 60 min, and rinsed again three times 10 min in PBS. The slides were immediately coverslipped using Vectashield (with or without DAPI, H-1200, H-1000; Vector Laboratories) as mounting medium.
The presence of dopamine (DA) and glutamate was detected by immunofluorescence concurrently or separately. The brain was immersed for 17–18 h in a 0.05 m Tris buffered 0.1% sodium metabisulfite and 0.8% NaCl (TBSM; pH 7.4) solution containing 2% glutaraldehyde. The brain was then transferred to TBSM containing 20% (w/v) sucrose overnight. The last two steps were performed at 4°C. The next day, the brain was cut on a cryostat. Sections of 25 μm thickness were collected on slides and air-dried overnight. The sections were then rinsed three times 10 min and incubated in TBSM containing 1% sodium borohydride for 30 min. After three rinses in TBSM, the sections were incubated in TBSM containing 5% normal goat serum and 0.3% Triton X-100 for 60 min. The sections were then incubated overnight at 4°C in the blocking solution containing the DA and/or the glutamate primary antibodies. The next day, the sections were rinsed three times 10 min with TBSM, incubated in the blocking solution containing the appropriate secondary antibodies (see below) for 60 min, and rinsed again three times 10 min in TBSM. The slides were then immediately coverslipped as described for TH.
For TH immunofluorescence, a rabbit anti-TH primary antibody was used (diluted 1:400; AB152; Millipore) followed by a donkey anti-rabbit AlexaFluor 594 (diluted 1:400; A21207; Invitrogen) or a goat anti-rabbit AlexaFluor 488 (1:400; A11008; Invitrogen). TH immunofluorescence was sometimes used in the present study as an alternative to DA immunofluorescence to visualize the DA neurons in the PT, based on studies in lampreys (Pombal et al., 1997; Abalo et al., 2005; Pierre et al., 1997; Ryczko et al., 2013). The AB152 antibody against TH has been used reliably on lampreys in many independent studies by different research groups on DA neurons (Villar-Cerviño et al., 2006; Barreiro-Iglesias et al., 2008; Robertson et al., 2012; Ryczko et al., 2013). For DA immunofluorescence, a mouse anti-DA primary antibody was used (diluted 1:400; MAB5300; Millipore) followed by a either a goat anti-mouse AlexaFluor 488 (diluted 1:400; A11001; Invitrogen) or a goat anti-mouse DyLight 594 (diluted 1:400; 115-515-146; Jackson ImmunoResearch). The specificity of the MAB5300 antibody for DA was tested by ELISA by the manufacturer, and its pattern of labeling in our material corresponded closely to that reported with other DA antibodies in the lamprey (Abalo et al., 2005; Pierre et al., 1997). We have also independently confirmed that the TH-immunoreactive neurons in the PT use DA as neurotransmitter (Ryczko et al., 2013). For glutamate immunofluorescence, a rabbit polyclonal primary antibody directed against glutamate was used (diluted 1:5000; IG1007, lot 3603, ImmunoSolution) followed by either a goat anti-rabbit AlexaFluor 594 antibody (diluted 1:400; A11012, Invitrogen) or a goat anti-rabbit AlexaFluor 488 (diluted 1:400; A11008, Invitrogen). The IG1007 glutamate antibody has been used successfully to label glutamatergic neurons in the lamprey brain (Barreiro-Iglesias et al., 2008; Villar-Cerviño et al., 2011; Fernández-López et al., 2012) and in salamanders (Ryczko et al., 2016a). The specificity of the antibody was confirmed by dot blots performed by the supplier, which revealed no immunoreaction against a variety of amino acid conjugates such as aspartate. Western blots did not yield staining of lamprey brain proteins extracts (Barreiro-Iglesias et al., 2008; Villar-Cerviño et al., 2011). The staining is similar to that obtained with a mouse monoclonal antibody directed against glutamate (Fernández-López et al., 2012). Brain regions stained by the antibody used in the present study also contained neurons expressing the vesicular transporter for glutamate mRNA (Villar-Cerviño et al., 2011).
In double DA/glutamate immunofluorescence experiments, the primary antibodies, and the secondary antibodies, were mixed together. The specific labeling obtained under these conditions was in every ways similar to the labeling obtained when the antibodies were each used separately. Biocytin was visualized with streptavidin Alexa Fluor594, 488, or 350 (diluted 1:400; Life Technologies), which was added to the secondary antibodies' solution without altering the immunofluorescence labeling.
In all cases, omitting the primary antibody from the procedures resulted in the absence of specific labeling on the brain sections.
The sections were then observed and photographed using an E600 epifluorescence microscope equipped with a DXM1200 digital camera (Nikon). For some sections, a confocal microscope was used (FV1000, Olympus). Photoshop CS5 (Adobe) was used to combine digital photomicrographs taken with different filter sets and to adjust the levels so that all fluorophores were clearly visible simultaneously.
Statistics.
Data in the text are presented as the mean ± SEM. For Ca2+ imaging, measurements of the area under the curve from the beginning of the response to the return to baseline were calculated using a calculation script in Clampfit (Molecular Devices). The values were expressed in ΔF/F × s. Correlations between variables and their significance as well as 95% confidence intervals were calculated using Sigma-Aldrich Plot 11.0. No statistical method was used to predetermine sample sizes. The sample sizes in the present study are in general similar to those used in the field. No randomization or blinding procedure was used. Parametric analyses were used when assumptions for normality and equal variance were respected, otherwise nonparametric analyses were used. Two-tailed paired Student's t tests were performed for comparing means between two dependent groups. For more than two dependent groups, a parametric one-way ANOVA for repeated measures or a nonparametric Friedman ANOVA on ranks for repeated measures was used. When two factors were tested, a two-way ANOVA for repeated measures on ranks was performed. Both ANOVA analyses were followed by a Student–Newman–Keuls post hoc test for multiple comparisons between groups. Statistical differences were assumed to be significant when p < 0.05.
Data availability.
All relevant data are available from the authors.
Results
We first confirmed that electrical stimulation of the PT robustly evokes swimming (Fig. 1A,B) in a lamprey semi-intact preparation in which the brain is exposed for stimulation, while the body swims in the chamber (Derjean et al., 2010; Gariépy et al., 2012a; Ryczko et al., 2013). We then examined whether PT stimulation could evoke graded locomotor output. A progressive increase in the PT stimulation intensity (1–32 μA, 4–5 Hz, 2 ms pulses, 10 s trains) elicited a graded increase in swimming frequency (Fig. 1B–D; n = 17 preparations). The recorded swimming frequencies (0.3–3.6 Hz) are in the range observed in freely moving lampreys (Islam and Zelenin, 2008). Data were pooled by expressing the locomotor frequency as a percentage of the maximal locomotor frequency. The PT stimulation intensity was expressed as a percentage of maximal stimulation intensity, that was defined when the locomotor response reached a plateau and did not grow further as the stimulation intensity was increased, as previously done in lampreys (Brocard et al., 2010) and salamanders (Ryczko et al., 2016a). The relationship between PT stimulation intensity and locomotor frequency followed a sigmoidal function, with swimming frequencies reaching a plateau at higher stimulation intensities (Fig. 1E; R2 = 0.69, p < 0.0001, n = 213 trials pooled 17 preparations). These findings establish that incremental PT stimulation evokes graded intensification of swimming frequency.
Then, we examined whether MLR cells are directly controlled in a graded fashion by descending inputs from the PT. We performed Ca2+ imaging of MLR cells in isolated brain preparations. MLR cells were retrogradely labeled with a Ca2+ indicator injected in the dendritic field of reticulospinal neurons. The dorsal part of the brain (i.e., the tectum) was removed to provide access to MLR cells for Ca2+ imaging (Fig. 2A). Reticulospinal cells were recorded extracellularly to monitor the activity of brainstem locomotor networks downstream the MLR. Stimulation of the PT (Fig. 2B,C) elicited Ca2+ responses in MLR cells together with reticulospinal activity (Fig. 2D,E). Increasing the intensity of the PT stimulation increased the size of the Ca2+ response in MLR cells (Fig. 2F,G). The relationship between normalized Ca2+ responses in the MLR and PT stimulation intensity also followed a sigmoidal function (Fig. 2H; R2 = 0.71, p < 0.0001, n = 56 trials, 10 cells pooled from 4 preparations). These results indicate that incremental increase in PT stimulation evokes graded intensification of MLR cell activation.
We then examined whether glutamatergic transmission was involved. Bath application of glutamatergic antagonists (CNQX 25 μm, AP5 100 μm) dramatically decreased the area of the Ca2+ responses evoked in MLR cells by trains of PT stimulation (reduced by 86.5 ± 5.1%, p < 0.05 vs control, Student–Newman–Keuls test, n = 12 cells pooled from 4 preparations; Fig. 2I,J). The responses returned to 76.4 ± 16.1% of control after wash out (p < 0.05 vs CNQX/AP5, Student–Newman–Keuls test; Fig. 2K). Similarly, bath applied glutamatergic antagonists also significantly decreased the responses evoked by single-pulse stimulation in MLR cells recorded either with Ca2+ imaging (reduced by 90.7 ± 23.1%, p < 0.01 vs control, paired t test, n = 3 cells; Fig. 2L) or patch-clamp recordings (reduced by 64.7 ± 1.6% in cell 1, p < 0.05 vs control, Student–Newman–Keuls test; reduced by 38.3 ± 4.1% in cell 2, p < 0.001 vs control, Student–Newman–Keuls test; Fig. 2M). The PT to MLR connection is likely monosynaptic as MLR cell responses followed high-frequency (20 Hz) stimulation doublets applied to the PT (n = 5 cells of 5 tested, data pooled from 3 preparations; Fig. 2N). Altogether, this shows that glutamatergic transmission is involved in the PT-evoked responses in MLR cells.
Next, we investigated whether the descending glutamatergic drive from the PT to the MLR was responsible for the activation of the locomotor circuits downstream the MLR. Electrical stimulation of the PT elicited reticulospinal discharges in isolated brain preparations. Microinjection in the MLR of the glutamatergic antagonists CNQX (1 mm) and AP5 (0.5 mm) dramatically decreased the discharge frequency elicited in reticulospinal neurons by electrical PT stimulation (reduced by 92.3 ± 4.9%, p < 0.001 vs control, Student–Newman–Keuls test, n = 15 trials pooled from 3 preparations; Fig. 3A–C). After wash out, the responses recovered with spiking frequency returning to 86.7 ± 12.8% of control (p < 0.001 vs CNQX/AP5, Student–Newman–Keuls test; Fig. 3D). Because electrical stimulation can recruit fibers of passage in addition to local cell bodies, electrical stimulation was replaced with chemical stimulation (Fig. 3A,E–G). Such chemical stimulation likely activates local cell bodies that are dopaminergic, glutamatergic and possibly from other types. Microinjections of glutamate (2.5 mm) onto the PT elicited discharges in reticulospinal cells similar to those evoked by electrical stimulation. Local microinjections of CNQX (1 mm) and AP5 (0.5 mm) in the MLR decreased the discharge frequency of reticulospinal neurons (reduced by 84.0 ± 4.2%, p < 0.05 vs control, Student–Newman–Keuls test, n = 20 trials pooled from 4 preparations). Recovery was obtained after wash out of the glutamatergic antagonists, with spiking frequency increasing back to 97.8 ± 15.4% of control (p < 0.05 vs CNQX/AP5, Student–Newman–Keuls test). In semi-intact preparations (Fig. 3H), microinjections of CNQX (1 mm) and AP5 (0.5 mm) in the MLR dramatically decreased the frequency of locomotor movements evoked by PT stimulation (reduced by 61.3 ± 10.3%, p < 0.001 vs control, Student-Newman–Keuls test; Fig. 3I,J). Recovery was obtained after wash out with the locomotor frequency returning to 96.4 ± 6.4% of control (p < 0.001 vs CNQX/AP5, Student–Newman–Keuls test, n = 18 trials per condition pooled from 3 preparations; Fig. 3K). Similar results were obtained in semi-intact preparations for the discharge frequency of reticulospinal neurons following microinjections of glutamatergic antagonists in the MLR (reduced by 93.0 ± 3.0%, p < 0.001 vs control, Student–Newman–Keuls test). These effects were also reversed after washout with frequency returning to 71.5 ± 7.6% of control (p < 0.001 vs CNQX/AP5, Student–Newman–Keuls test). Altogether, these results show that the descending glutamatergic pathway from the PT to the MLR plays an important role in the activation of locomotor circuits downstream the MLR.
We then looked for the anatomical substrate of these effects. We examined whether glutamatergic PT neurons send a descending projection to the MLR with anatomical techniques. Glutamatergic neurons were found in the PT (n = 9 preparations). We injected an axonal tracer in the MLR to label retrogradely neurons in the PT (Fig. 4A,B) and we observed that several of those were immunoreactive for glutamate (Fig. 4C–I; n = 7 preparations). Because a descending dopaminergic pathway originating from the PT to the MLR was previously reported (Ryczko et al., 2013; Pérez-Fernández et al., 2014), we examined whether PT glutamatergic cells were a separate population from the PT dopaminergic ones. Using double-labeling experiments, we found that several PT neurons were colabeled for dopamine and glutamate (Fig. 5A–I; n = 6 preparations). To determine whether PT neurons co-storing dopamine and glutamate sent descending projections to the MLR, we added a retrograde tracer injection in the MLR (Fig. 6A,B) in a series of triple labeling experiments. We found that several PT cells projecting down to the MLR were colabeled for dopamine and glutamate (n = 4 preparations; Fig. 6C–K). PT neurons storing only dopamine or glutamate were also found to send descending projections to the MLR. Altogether, these results provide direct evidence of a glutamatergic pathway from the PT to the MLR, in parallel with the previously characterized dopaminergic pathway (Ryczko et al., 2013; Pérez-Fernández et al., 2014). There is thus an anatomical substrate for interaction between these two transmitters at the level of the MLR.
Next, we examined the role of the interactions between glutamatergic and dopaminergic inputs from the PT to the MLR in the graded control of locomotor movements. We previously showed that PT-evoked dopamine release in the MLR increases swimming frequency through the activation of D1 receptors (Ryczko et al., 2013). Here we examined in semi-intact preparations whether the activation of D1 receptors in the MLR is needed to evoke graded swimming when stimulating the PT with increasing intensities. When the D1 antagonist SCH 23390 (0.5 mm) was microinjected in the MLR, the locomotor frequency of PT-evoked swimming was decreased as expected, but graded control of swimming was still possible when stimulating the PT with increasing intensities (Fig. 7A). Similar observations were made when pooling the data by expressing stimulation intensity and locomotor frequency as a function of their maximal values (Fig. 7B–D). In control conditions, the relationship between stimulation intensity and swimming frequency followed a sigmoid function (R2 = 0.78, p < 0.0001, n = 81 trials pooled from 6 preparations) that persisted in the presence of the D1 antagonist injected in the MLR (R2 = 0.76, p < 0.0001, n = 81 trials pooled from 6 preparations), indicating that progressive increase in stimulation intensity still elicited graded increase in swimming. After wash out of the drug, the sigmoid relation was still present (R2 = 0.83, p < 0.0001, n = 81 trials pooled from 6 preparations). Statistical analysis confirmed that the D1 antagonist reversibly decreased locomotor frequency especially at maximal PT stimulation intensities (Fig. 7E). This might be explained by a “floor effect” at lower intensities, where the range of evoked swimming frequencies could be too narrow to detect significant differences between drug conditions. In the presence of the D1 antagonist, the locomotor frequencies reached a plateau that was lower than under control conditions at 50, 60, 70, 80, 90, and 100% of maximal stimulation intensities (p < 0.05 to p < 0.001 against control, Student–Newman–Keuls test after a two-way ANOVA on ranks for repeated measures, p < 0.001; Fig. 7E). This effect was abolished after wash out at 50, 60, 70, 90, and 100% of maximal stimulation intensities (p < 0.05 to p < 0.001 against drug injection, Student–Newman–Keuls test). Because we previously demonstrated that a small number of dopamine cells in the PT send descending projections to the contralateral MLR (Ryczko et al., 2013), microinjections of the D1 antagonist were performed on both sides of the MLR (n = 5 preparations). The effects were very similar to those observed with a unilateral injection of the drug. The swimming frequency was reduced at 40, 50, 60, 70, 80, 90, and 100% of maximal stimulation intensities (p < 0.05 to p < 0.001 against control, Student–Newman–Keuls test after a two-way ANOVA on ranks for repeated measures, p < 0.001) and graded control of locomotor frequency was still present with increasing intensities of PT stimulation (sigmoid fit: R2 = 0.62, p < 0.0001, n = 70 trials pooled from 5 preparations). Altogether, these results suggest that glutamatergic inputs from the PT to the MLR are necessary for evoking locomotor output in a graded fashion, whereas the dopaminergic inputs amplify the descending glutamatergic command through D1 receptors without contributing to the rheostat-like effect.
Discussion
Results from the present study show in lamprey that the PT, a structure homologous to the mammalian SNc, sends a descending glutamatergic input that controls MLR cell activity and swimming in a graded fashion. We found PT neurons labeled for both dopamine and glutamate that projected to the MLR, indicating a close interaction between these transmitters in the generation of the locomotor command. Blockade of D1 receptors in the MLR resulted in a reduced frequency of swimming movements without disrupting the graded control of locomotion by the PT, whereas blockade of glutamatergic receptors in the MLR nearly abolished locomotor output.
To our knowledge, results from the present study provide the first demonstration that the MLR can be activated incrementally by a higher brain region to control the speed of the locomotor movements. The substrate underlying this phenomenon consists of glutamatergic and dopaminergic inputs acting in parallel. We now show that the glutamatergic input is essential to elicit MLR activity and locomotion in a graded fashion, whereas the dopaminergic one provides additional excitation, but is not essential to evoke locomotion. In the presence of a D1 blocker, the relationship between stimulation intensity and swimming frequency was shifted to the right, indicating that a stronger stimulus was necessary to generate the same swimming frequency. This is consistent with our previous observation that dopamine application over the brainstem reduced the threshold of the current applied to the PT needed to elicit swimming (Ryczko et al., 2013). This suggests that D1 activation either increases the strength of glutamatergic inputs to the MLR or increases MLR cell excitability through modulation of their intrinsic properties.
We found that some PT cells projecting to the MLR colocalize dopamine and glutamate. There is accumulating evidence that neurons with multiple transmitters are present in several neural circuits from lamprey to mammals (for review, see Granger et al., 2017). In zebrafish, catecholaminergic neurons were also found to coexpress either glutamate or GABA markers, including in the PT, where dopaminergic neurons preferentially use glutamate as a second transmitter (Filippi et al., 2014). In mammals, neurons co-storing dopamine and glutamate were found in the SNc and ventral tegmental area (Sulzer et al., 1998; Chuhma et al., 2004; Dal Bo et al., 2004; Kawano et al., 2006; Hnasko et al., 2010; Stuber et al., 2010; Fortin et al., 2012; for review, see Morales and Root, 2014). A future avenue would consist in determining whether the release of each transmitter is activity-dependent and targets different cell populations within the MLR.
The mechanisms underlying the interplay between dopaminergic and glutamatergic inputs from the primal SNc in the graded activation of MLR cells remain to be investigated. In mammals, D1 receptor activation increases AMPA (Cepeda et al., 1993; Hernández-Echeagaray et al., 2004; Tseng and O'Donnell, 2004; Han and Whelan, 2009) and NMDA-mediated responses (Cepeda et al., 1993; Tseng and O'Donnell, 2004). D2 receptors are also present in lampreys (Robertson et al., 2012; Ericsson et al., 2013; Pérez-Fernández et al., 2014). In mammals D2 receptor activation decreases both AMPA- and NMDA-mediated components of glutamatergic responses (Cepeda et al., 1993; Hernández-Echeagaray et al., 2004; Tseng and O'Donnell, 2004). Dopamine could also modulate glutamatergic transmission presynaptically as shown in mammals (for review, see Tritsch and Sabatini, 2012) and lamprey spinal cord (Svensson et al., 2003a,b). There is also a possibility that dopamine acts directly on MLR cell intrinsic properties as shown in the lamprey striatum (Ericsson et al., 2013).
It is noteworthy that other forebrain regions send input to brainstem motor networks. The pallium (homologous to the mammalian cortex) sends projections to the MLR and reticulospinal neurons (Ocaña et al., 2015). Pallium stimulation elicits reticulospinal responses and sometimes elicits swimming in lampreys (Ocaña et al., 2015), but whether incremental pallium activation can control MLR cell responses and locomotor movements is not known. In mammals, motor cortex stimulation is rather linked to voluntary modification of limb trajectory during ongoing locomotion (Bretzner and Drew, 2005). Moreover, another locomotor center, called the diencephalic locomotor region (considered to be equivalent to the subthalamic locomotor region in mammals), sends direct input to reticulospinal neurons (El Manira et al., 1997; Ménard and Grillner, 2008). The possibility that our present results might be explained by the recruitment of descending fibers originating from these two regions is unlikely, because the activation of the MLR was also obtained when stimulating the PT with local injections of glutamate, thus circumventing axonal recruitment. Future studies should seek to determine how these different inputs from supra-MLR regions are integrated by brainstem motor centers to shape motor output.
At the functional level, the dual inputs could tune the activity of MLR cells for the animal to appropriately approach targets or avoid threats, an essential function for survival. In this context, it is noteworthy that the olfactory bulb sends direct inputs to the PT (Derjean et al., 2010). Olfactory information could thus generate different levels of PT activity to control locomotor output in order for the animal to reach or avoid olfactory targets. Future studies should examine whether and how olfactory information controls the activity of cell populations in the PT.
The lamprey is considered as a blueprint of the vertebrate brain (Robertson et al., 2014), suggesting that the descending glutamatergic pathway from the SNc to brainstem locomotor networks reported here could be present in all vertebrates. In line with this possibility, we recently provided evidence that the descending dopaminergic pathway from the SNc to the MLR is conserved in amphibians, mammals, and likely humans (Ryczko et al., 2016b; for review, see Ryczko and Dubuc, 2017). Furthermore, the present study suggests that a dysfunction in the regulation of the interaction between glutamate and dopamine in the MLR could translate into abnormal activation of the locomotor system, and thus result in locomotor deficits such as those reported in Parkinson's disease and other locomotor disorders.
Footnotes
This work was supported by the Canadian Institutes of Health Research (15129); the Fonds de la Recherche en Santé du Québec (Groupe de Recherche sur le Système Nerveux Central, GRSNC, 5249); the Natural Sciences and Engineering Research Council of Canada Grant 217435 (R.D.); the Great Lakes Fishery Commission Grants 54011 and 54021 (R.D.); fellowships from the Fonds de la Recherche en Santé du Québec (D.R.) and the GRSNC Jasper fellowship (D.R.); and scholarships from the University of Cologne and the German Academic Exchange Service (S.G.). We thank Arlette Kolta for providing access to her electrophysiology setup for some experiments, Danielle Veilleux for her technical assistance, and Frédéric Bernard for his help with the graphics.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Réjean Dubuc, Groupe de Recherche en Activité Physique Adaptée, Département des Sciences de l'Activité Physique, Université du Québec à Montréal, C.P. 8888, Section Centre-ville, Montréal, QC H3C 3P8, Canada. rejean.dubuc{at}gmail.com