Abstract
The transfer of retinal signals from the dorsal lateral geniculate nucleus (dLGN) to the primary visual cortex (V1) is modulated by a variety of extraretinal inputs, including extrinsic connections formed by GABAergic neurons in the pretectum (PT) and visual sector of the thalamic reticular nucleus (vTRN), as well as the intrinsic connections of GABAergic dLGN interneurons. In the current study, we determined how GABAergic PT projections to the dLGN and vTRN can influence retinogeniculate transfer using a variety of viral tracing techniques, electron microscopy, in vitro physiological recordings, and optogenetics in male and female mice. We found that the PT provides over 75% of the GABAergic and over 30% of the total synaptic input to the vTRN. Optogenetic activation of PT terminals reduced the firing frequency of vTRN neurons as well as the amplitudes of their postsynaptic responses to V1 input. In the dLGN, synaptic terminals originating from the PT targeted interneurons more frequently than thalamocortical (relay) cells, and optogenetic activation of PT input had a greater impact on interneuron firing frequency compared with relay cells. This cell type-specific impact of PT input to the dLGN resulted in the disinhibition of relay cells and an increase in the amplitude of their postsynaptic responses to retinal input. Taken together, our results indicate that GABAergic PT projections to the visual thalamus serve to boost retinogeniculate transfer via two types of disinhibition, potentially enhancing the flow of visual information to V1 following gaze shifts.
- dorsal lateral geniculate nucleus
- interneuron
- optogenetics
- thalamic reticular nucleus
- ultrastructure
- vision
Significance Statement
The transfer of visual information from the retina to the cortex must be coordinated with gaze shifts to explore the surrounding environment. Here we document pathways from the pretectum, a region responsive during gaze shifts, to the visual thalamus. We find that pretectum neurons that use the inhibitory neurotransmitter gamma amino butyric acid (GABA) selectively innervate GABAergic neurons in the thalamic reticular nucleus and dorsal lateral geniculate nucleus, which serves to boost the transfer of visual information from the thalamus to the cortex. The identification of these brain circuits may have important implications for disorders of gaze shifts and/or sensory modulation.
Introduction
The activity of the dorsal lateral geniculate nucleus (dLGN) is sculpted by a variety of circuits that utilize GABA as a neurotransmitter (Wang et al., 2011a; Hirsch et al., 2015), but it is unclear how these connections collaborate to shape visual signals. Extrinsic sources of GABAergic input include the thalamic reticular nucleus (TRN) and pretectum (Cucchiaro et al., 1991; Wahle et al., 1994; Wang et al., 2001, 2002; Soto-Sánchez et al., 2017; Campbell et al., 2020) while intrinsic inhibition is mediated by GABAergic interneurons which can provide both local and global inhibition via presynaptic dendrites (Acuna-Goycolea et al., 2008; Casale and McCormick, 2011; Wang et al., 2011b; Cox and Beatty, 2017; Morgan and Lichtman, 2020; Jager et al., 2021; Maher et al., 2023). To fully understand how dLGN activity is shaped by inhibition, it is important to determine how these the sources of GABAergic input interact both inside and outside this nucleus.
The TRN is composed of a thin sheet of GABAergic neurons that surround and reciprocally connect with the dorsal thalamus (Guillery et al., 1998; Guillery and Harting, 2003). The TRN is divided into modality-specific sectors (Martinez-Garcia et al., 2020; Sokhadze et al., 2022; Campbell et al., 2024) which have been proposed to create a “searchlight” of attentional selection by boosting the activity of thalamocortical neurons representing attended targets and inhibiting those representing unattended targets (Crick, 1984). Indeed, in the mouse dLGN, GABAergic input from the visual sector of the TRN (vTRN) specifically targets thalamocortical (relay) cells and can strongly inhibit their activity to control the gain of sensory transmission during divided attention tasks (Wimmer et al., 2015; Campbell et al., 2020).
The PT is a large area between the superior colliculus and the thalamus which is involved in a variety of sensory-motor behaviors. In the cat, PT neurons that project to the dLGN are GABAergic and are designated “saccade neurons” based on their responses to fast image displacements and during the execution of saccadic eye movements in both the light and the dark (Cucchiaro et al., 1991; Wahle et al., 1994; Schmidt, 1996). Input from PT “saccade” neurons has been shown to exclusively target GABAergic interneurons in the A lamina of the cat dLGN (Cucchiaro et al., 1993; Wang et al., 2002) and is therefore proposed to disinhibit relay cells to increase the gain of sensory transmission following the acquisition of visual targets via saccadic eye movements (Schmidt, 1996; Wang et al., 2002). The cat pretectum also innervates the vTRN (Uhlrich and Manning, 1995), potentially providing a second circuit to disinhibit dLGN relay cells.
The mouse dLGN provides an excellent model for untangling the connections formed by GABAergic elements. In this species, we used a variety of anatomical and in vitro optogenetic techniques to investigate the synaptic properties of connections formed by PT terminals in both the dLGN and vTRN. Collectively, our results suggest that GABAergic PT projections provide the circuitry needed to create a “searchlight” of visual attention by boosting geniculate relay cell activity via disinhibition.
Materials and Methods
Animals
All breeding and experimental procedures were approved by the University of Louisville Institutional Animal Care and Use Committee. Experiments were carried out using mice, of either sex, of a C57BL/6 line (Jackson Labs stock number 000664), a line in which neurons that contain the 65KD isoform of glutamic acid decarboxylase (GAD) express green fluorescent protein (GAD65-GFP; López-Bendito et al., 2004), a line in which neurons that contain the 67KD isoform of GAD express green fluorescent protein (GFP67-GFP; Jax Stock No: 007677, G42 line; López-Bendito et al., 2004), a GAD2-cre driver line (Gad2tm2(cre)Zjh/J, Jackson Labs stock number 010802), a line in which layer 6 cortical neurons express cre-recombinase (“L6-cre”, Ntsr1-GN220; RRID:MMRRC 017266-UCD), or an Ai9 reporter line [B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J, Jackson Labs stock number 007909; Table 1].
The experiments, animal lines, animal numbers, analyses, software, and statistical tests used in this study
Herpes simplex virus injections
To label neurons that project to the dorsal lateral geniculate nucleus (dLGN) or visual sector of the thalamic reticular nucleus (vTRN) via retrograde transport, adult GAD2-cre or Ai9 mice were deeply anesthetized with a mixture of ketamine (100–150 mg/kg) and xylazine (1–1.5 mg/kg). The analgesic meloxicam (20 mg/kg) was also injected prior to surgery, and the anesthetic bupivacaine (3 g/kg) was injected in the skin of the scalp. The eyes were covered with a lubricant ointment, and the animals were then placed in a stereotaxic apparatus (Angle Two Stereotaxic, Leica). An incision was made along the scalp, a small hole was drilled in the skull, and a 34-gauge needle was lowered into the dLGN (2.3 mm caudal, 2 mm lateral and 2.75 mm ventral to Bregma) or vTRN (1.68 mm caudal, 2 mm lateral, and 2.85 mm ventral to bregma). The needle was attached to a NanoFil syringe inserted in an ultramicropump, and the pump was used to deliver volumes of 50–150 nl at a rate of 20 nl/min of heF1α-LS1L-mCherry HT (retrograde HSV-1, packaged via amplicon system and purified on a sucrose gradient, titer = 5 × 109 infectious units/ml) in GAD2-cre mice to induce the expression of mCherry in a cre-dependent manner in retrogradely labeled neurons or heF1α-EYFP-IRES-cre HT (retrograde HSV-1, packaged via amplicon system and purified on a sucrose gradient, titer = 5 × 109 infectious units/ml) in Ai9 mice to induce cre-recombinase, EYFP, and tdTomato in retrogradely labeled neurons (both viruses purchased from the MIT viral vector core, now available at UNC viral vector core, RN411 and RN405, respectively).
After removal of the needle, the scalp skin was sealed with tissue adhesive (n-butyl cyanoacrylate), and the animals were placed on a heating pad until mobile. Post-surgery, animals were carefully monitored for proper wound healing, and meloxicam (20 mg/kg) was administered once per day for 48 h. Mice were killed for histology between 7 and 14 d post injection, as described below.
Adeno-associated virus (AAV) injections for anatomy
To label GABAergic axon projections, Cre-dependent viruses, Flex-rev-oChIEF-tdTomato (plasmid #30541, Addgene, packaged using AAV serotype 2/1, titer = 2.0 × 1012 vg/ml), CAG-FLEX-EGFP-WPRE.bGH (plasmid #51502, Addgene, packaged using AAV serotype 1, titer ≥ 1 × 1013 vg/ml), hSyn-DIO-EGFP (plasmid #50457, Addgene, packaged using AAV serotype 1, titer ≥ 7 × 1012 vg/ml), and Syn-FLEX-rc[ChrimsonR-tdTomato] (plasmid #62723, Addgene, packaged using AAV serotype 5, titer ≥ 5 × 1012 vg/ml) were injected unilaterally into the pretectum (PT), hypothalamus/ventral thalamus, and/or globus pallidus. For virus delivery, adult male or female GAD2-cre mice were anesthetized and placed in a stereotaxic apparatus as described above. An incision was made along the scalp, and a small hole was created in the skull above the PT (from bregma 2.68–2.8 mm posterior, 0.5–0.65 mm lateral, and 2.2–2.3 mm ventral), hypothalamus/ventral thalamus (from bregma: 1.54–1.64 mm posterior, 1.21–1.0 mm lateral, 4.84 mm ventral), and/or globus pallidus (from bregma: 0.94 mm posterior, 2.37 mm lateral, 3.78 mm ventral). Virus was delivered via a 34-gauge needle attached to a NanoFil syringe inserted in an ultramicropump. Volumes of 50–100 nl were injected at each site at a rate of 20–30 nl/min.
To label the projections of GFP-containing cells in the PT of GAD67-GFP mice, two GFP-dependent viruses (pAAV1-Ef1a-C-CreinG, Addgene #69571-AAV1, titer ≥ 1 × 1013 vg/ml and pAAV1-Ef1a-N-CretrcintG, Addgene #69570-AAV1, titer ≥ 1 × 1013 vg/ml) were combined in one syringe in a 1:1 ratio and injected simultaneously in the PT via a 34-gauge needle attached to a NanoFil syringe inserted in an ultramicropump (100–150 nl at a rate of 30 nl/min). This was immediately followed by an injection in the same site of a cre-dependent virus to express mCherry (DIO-mCherry-WPRE, packaged using AAV serotype 8.2, MIT viral vector core AAV-RN12, 2.13 × 1013 vg/ml) or tdTom (Syn-FLEX-rc[ChrimsonR-tdTomato], Addgene #62723-AAV5, titer ≥5 × 1012 vg/ml, 100 nl at a rate of 30 nl/min).
To label the projections of layer 6 V1 cortical neurons, a virus that induces the expression of APEX2 throughout the cytoplasm in a cre-dependent manner (AAV-Ef1a-DIO-dAPEX2, packaged using AAV serotype 9, Addgene plasmid #117174, gift from Dr. David Ginty) was injected unilaterally in V1 of adult L6-cre mice. For virus delivery, mice were anesthetized and placed in a stereotaxic apparatus as described above. An incision was made along the scalp, and a small hole was created in the skull above V1 (from bregma 3.40 mm posterior, 2.20 mm lateral, and 1.29 mm ventral). Virus was delivered via a 34-gauge needle attached to a NanoFil syringe inserted in an ultramicropump. A volume of 150 nl was injected at a rate of 30 nl/min. For all AAV injections, after removal of the needle, the scalp skin was sealed, and the animals were treated as described above.
Cholera toxin subunit B and biotinylated dextran amine injections
To delineate the vTRN, adult male and female GAD67-GFP mice were anesthetized and placed in a stereotaxic apparatus as described above. An incision was made along the scalp, and a small hole was drilled in the skull above the dLGN or pulvinar nucleus. A glass pipette (20–40 µm tip diameter) containing a 0.2% solution of cholera toxin subunit B conjugated to Alexa Fluor 546 (CTB-546) in phosphate-buffered saline (PBS; 0.01 M phosphate buffer with 0.9% NaCl) was lowered into the dLGN (from bregma: 2.14 mm posterior, 2.0 mm lateral, 2.89 mm ventral) or pulvinar nucleus (from bregma: 2.11 mm posterior, 1.73 mm lateral, 2.73 mm ventral), and CTB was iontophoretically ejected using 3 µA continuous positive current for 15–30 min. After removal of the pipette, the scalp skin was sealed, and the animals were treated as described above.
To label PT projections via the anterograde transport of BDA, C57BL/6J mice were deeply anesthetized and placed in a stereotaxic apparatus, and a small hole in the skull overlying the PT was prepared as described above for virus injections. A glass pipette (20–40 μm tip diameter) containing a 5% solution of biotinylated dextran amine (BDA, Molecular Probes) in saline was lowered into the PT, and BDA was iontophoretically ejected using 3 μA continuous positive current for 20 min. After removal of the pipette, the scalp skin was sealed, and the animals treated and monitored as described above.
Histology of tissue used for anatomical analyses
Two days to two weeks following injection of tracers and/or viruses, mice were deeply anesthetized with Avertin (0.5 mg/g) or ketamine (450 mg/kg) and transcardially perfused with a fixative solution of 4% paraformaldehyde or freshly prepared 2% paraformaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4 (PB). Additional GAD67-GFP and GAD65-GFP mice that were not injected were also perfused for immunocytochemistry. In each case, the brain was removed from the skull, and 70-µm-thick coronal sections were cut using a vibratome (Leica Microsystems). Sections that contained fluorescent labels were mounted on slides and imaged using a confocal microscope (Olympus FV1200BX61) or additionally stained using antibodies as described below.
Selected sections were incubated overnight in either a parvalbumin antibody (made in mouse, Sigma, catalog #P3088, 1:2,000; RRID:AB_477329), a vGAT antibody (made in rabbit, Synaptic Systems, catalog #131 103; RRID: AB_887870), a NeuN antibody (made in mouse, Chemicon, catalog # MAB377 RRID:AB_2298772, or made in rabbit Millipore, catalog #ABN78; RRID:AB_10807945), or an antibody against the type 2 vesicular glutamate transporter (vGLUT2; made in guinea pig, Millipore Sigma, catalog #AB2251, RRID:AB_2665454). The following day, the sections were incubated in a 1:100 dilution of goat-anti-mouse, goat-anti-rabbit, and/or goat anti-guinea pig antibodies that were directly conjugated to fluorescent compounds (Alexa Fluor 488, 546 or 633; Invitrogen). The sections were then mounted on slides and imaged using a confocal microscope (Olympus FV1200BX61).
To label tissue for viewing in a transmitted light microscope or transmission electron microscope, sections that contained GFP were incubated overnight in a rabbit anti-GFP antibody (Millipore, catalog #AB3080, RRID:AB_91337, 1:1,000) and subsequently incubated for 1 h in a 1:100 dilution of a biotinylated goat-anti-rabbit antibody (BA100, Invitrogen, RRID: AB_2313606), followed by avidin and biotinylated horseradish peroxidase (Vectastain Elite, ABC-HRP Kit, Vector Laboratories, RRID: AB2336819, 1 h) and reacted with nickel-enhanced diaminobenzidine (DAB). Sections that contained BDA were incubated in ABC-HRP and reacted with nickel-enhanced DAB. Sections that contained APEX2 were reacted with nickel-enhanced DAB. DAB-stained sections were then mounted on slides and imaged using transmitted light or processed for electron microscopy as described below.
Electron microscopy
Sections that contained terminals labeled by the anterograde transport of BDA, virus expression of APEX2, or the GFP antibody were postfixed in 2% osmium tetroxide, dehydrated in an ethyl alcohol series, and flat embedded in Durcupan resin between two sheets of Aclar plastic (Ladd Research). Durcupan-embedded sections were first examined with a light microscope to select areas for electron microscopic analysis. Selected areas were mounted on blocks, ultrathin sections (70–80 nm, silver-gray interference color) were cut using a diamond knife and collected on Formvar-coated nickel slot grids. Selected sections were stained for the presence of GABA. A postembedding immunocytochemical protocol described previously (Bickford et al., 2010, 2015) was employed. Briefly, we used a 0.25 μg/ml concentration of a rabbit polyclonal antibody against GABA (Sigma-Aldrich, catalog #A2052, RRID: AB_477652), and the GABA antibody was tagged with a goat-anti-rabbit antibody conjugated to 15 nm gold particles (BBI Solutions USA). The sections were air dried and stained with a 10% solution of uranyl acetate in methanol for 30 min before examination with an electron microscope (Hitachi HT 7700).
For ultrastructural analysis, in sampled single sections all BDA-, APEX2-, GFP-, and/or GABA-labeled terminals involved in a synapse were imaged. The pre- and postsynaptic profiles were characterized based on size (cross-sectional area at the location of the synapse, measured using Image J, RRID: nif-000-30467), the presence or absence of synaptic vesicles, and overlying gold particle density. For presentation of ultrastructural features, electron microscopic images were imported into Adobe Photoshop software, and the brightness and contrast were adjusted.
Adeno-associated virus injections for in vitro physiology
To label and/or activate PT projections, unilateral or bilateral AAV injections were placed in the PT. For virus delivery, P22–49 GAD2-cre or GAD67-GFP mice were anesthetized and placed in a stereotaxic device as described above. An incision was made along the scalp, and a small hole was drilled in the skull above the left and/or right PT (2.68–2.8 mm caudal, 0.5–0.65 mm lateral and 2.2–2.3 mm ventral to bregma). Virus was delivered via a 34-gauge needle attached to a NanoFil syringe inserted in an ultramicropump. Volumes of 75–200 nl were injected at each site at a rate of 20–50 nl/min. The following AAVs were injected into the PT: AAV serotype 2/1 carrying a vector for the Channelrhodopsin variant Chimera EF with I170 mutation (ChIEF) fused to the red fluorescent protein, tdTomato, AAV 2/1 Flex Chief-TdTomato (production details in Jurgens et al., 2012), AAV9-Ef1a-double floxed-hChR2 (H134R)-EYFP-WPRE-HGHpA (Addgene plasmid #20298), or pAAV9-Syn-ChrimsonR-TdTomato (Addgene plasmid #59171). After removal of the needle, the scalp skin was sealed, the animals treated and monitored as described above, and 11–29 d after surgery used for in vitro physiology as described below.
To label and activate both PT and retinogeniculate projections, GAD2-cre pups received unilateral intravitreal injections of pAAV9-Syn-ChrimsonR-tdT (Addgene plasmid #59171) shortly after eye opening [approximately at postnatal day (P) 14]. Each pup was anesthetized with isoflurane (1–1.5% via a small nose cone), the sclera was pierced with a sharp-tipped glass pipette, and excess vitreous was drained. Another pipette filled with the AAV solution was inserted into the hole made by the first pipette. The pipette containing the AAV was attached to a picospritzer and a volume of ∼1 μl of solution was injected into the eye. The nose cone used to administer isoflurane was then removed, and, once alert, the pup was returned to the cage containing the dam and littermates. Approximately 1 week after eye injections (after weaning), these mice received bilateral AAV injections in the PT as described above, using AAV9-Ef1a-double floxed-hChR2 (H134R)-EYFP-WPRE-HGHpA (Addgene plasmid #20298). Ten to 32 d after the PT injections, these mice were used for in vitro physiological experiments as described below.
To label and activate both PT and visual cortex (V1) projections, unilateral or bilateral V1 injections of pAAV9-Syn-ChrimsonR-tdT (Addgene plasmid #59171) were paired with unilateral or bilateral PT injections of AAV9-hSyn-flex-Chronos-GFP (Addgene plasmid #59170) in P22–60 GAD2-cre mice. For virus delivery, mice were anesthetized and placed in a stereotaxic device as described above. An incision was made along the scalp, and a small hole was drilled in the skull above the left and/or right V1 (3.55–3.88 mm caudal, 2.04–2.6 mm lateral and 1.07–1.57 mm ventral to Bregma) and PT (2.8 mm caudal, 0.5–0.84 mm lateral and 2.31–2.41 mm ventral to Bregma). Virus was delivered via a 34-gauge needle attached to a NanoFil syringe inserted in an ultramicropump. Volumes of 75–200 nl were injected at each site at a rate of 20 nl/min. After removal of the needle, the scalp skin was sealed, and the animals were treated and monitored as described above and used for in vitro physiology as described below.
Slice preparation and optogenetic stimulation
Mice used for slice preparation ranged in age from P33 to P75 (average age P46.5). Ten to 14 d following PT virus injections, PT and intravitreal virus injections, or PT and V1 injections, mice were deeply anesthetized with isoflurane (4–5%) and decapitated. The brain was removed from the head, chilled in cold slicing solution (in mM: 2.5 KCl, 26 NaHCO3, 1.25 NaH2PO4, 10 MgCl2, 2 CaCl2, 234 sucrose, and 11 glucose) for 2 min and quickly transferred into a Petri dish with room temperature slicing solution to block the brain for subsequent sectioning. Coronal slices (300 μm) were cut in cold slicing solution using a vibratome (Leica VT1000 S). Slices were then transferred into a room temperature incubation solution of oxygenated (95% O2/5% CO2) artificial cerebrospinal fluid (ACSF) containing the following (in mM: 126 NaCl, 26 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 2 MgCl2, 2 CaCl2, and 10 glucose) for 30 min to 6 h. Individual slices were transferred into a recording chamber, which was maintained at 32°C by an inline heater and continuously perfused with room temperature oxygenated ACSF (2.5 ml/min, 95%O2/5%CO2). Slices were stabilized by a slice anchor or harp (Warner Instruments 64-0252). Neurons were visualized on an upright microscope (Olympus BX51WI) equipped with both differential interference contrast optics and filter sets for visualizing CTB-488 and YFP (Chroma 49002) or tdTomato (Chroma 49005) using a 4× or 60× water-immersion objective (Olympus) and a CCD camera.
Recording electrodes were pulled from borosilicate glass capillaries (World Precision Instrument) by using a Model P-97 puller (Sutter Instrument). The electrode tip resistance was 4–6 MΩ when filled with either of the two internal solutions used for recording. For current-clamp recordings, the internal solution was a K+-based solution that contained the following components (in mM): 117 K-gluconate, 13.0 KCl, 1 MgCl2, 0.07 CaCl2, 0.1 EGTA, 10 HEPES, 2 Na2-ATP, and 0.4 Na2-GTP with pH adjusted to 7.3 with KOH and osmolarity 290–295 mOsm. For voltage-clamp recordings at −60 and/or 0 mV, the internal solution was a Cs+-based solution that contained the following components (in mM): 117 gluconic acid, 10.0 CsCl, 1 MgCl2, 0.07 CaCl2, 0.1 EGTA, 10 HEPES, 2 Na2-ATP, and 0.4 Na2-GTP with pH adjusted to 7.3 with CsOH and osmolarity 290–295 mOsm. Biocytin (0.5%) was also added to these solutions to allow morphological reconstruction of the recorded neurons.
Whole-cell recordings were obtained from the dLGN or vTRN. For PT injection experiments in GAD2-cre mice, cells within the PT termination zones were targeted for recording. For PT injection experiments in GAD67-GFP mice, GFP-positive or GFP-negative dLGN cells within the PT termination zones were targeted for recording. For PT and intravitreal injection experiments in GAD2-cre mice, dLGN cells within the PT and ipsilateral or contralateral retinogeniculate termination zones were targeted for recording. For PT and V1 injection experiments in GAD2-cre mice, vTRN cells within the PT and V1 termination zones were targeted for recording. Video images of the patched cell locations, and the presence or absence of GFP within patched cells, were recorded using the CCD camera.
Recordings were obtained with an Axon Instruments MultiClamp 700B amplifier (Molecular Devices), and a Digidata 1440A was used to acquire electrophysiological signals. The stimulation trigger was controlled by Clampex 11.03 software (Molecular Devices). The signals were sampled at 20 kHz, and data were analyzed offline using Matlab. For photoactivation of PT, V1, and/or retinogeniculate terminals, light from a blue (Prizmatix UHP 460) and/or red (Prizmatix UHP 630) light-emitting diode was reflected into a 60× water-immersion objective. This produced spots of light onto the submerged slice with diameters of ∼0.3 mm (at full power, blue light 107.2 mW/mm2, red light 228.7 mW/mm2). Pulse duration and frequency were under computer control. For repetitive stimulation of PT terminals alone, pulse durations of 1 ms were used at frequencies of 1, 2, 5, 10, 20, and 50 Hz. To block GABAergic transmission pharmacologically, in some experiments GABA receptors (GABAA) were blocked via bath application of the antagonist 2-(3-carboxypropyl)-3-amino-6-(4-methoxyphenyl)-pyridazinium bromide (SR95531, 20 μM; Tocris Bioscience, catalog #1262).
To examine how photoactivation of PT terminals affects spiking activity, neurons were held at 5–10 mV below firing threshold and then given a 1.5 s square wave depolarizing current pulse of sufficient strength to evoke a steady train of spike firing >5 Hz. Changes in spike firing were then calculated by comparing equivalent periods (0.5 s) of activity in the presence or absence of blue or red light stimulation (2, 5, 10, 20, and 50 Hz). To minimize the effects of frequency accommodation or burst firing, measures of spike firing were obtained 0.5 s after the onset of the 1.5 s current pulse. Typically, the measurements described above were based on the average of three stimulus presentations. For interneuron and TRN cells that did not display regular spiking characteristics, this protocol was modified. For these cells, a 500 ms square wave depolarizing current pulse began 100 ms after the start of a 500 ms 50 Hz train of light pulses.
For the simultaneous activation of V1 and PT terminals, the following light activation protocol was used: single (1 ms duration) pulses of red light, followed by single pulses of red light paired with 400 ms of 50 Hz trains of blue light pulses (1 ms duration). For activation of the optic tract (OT), a square-wave pulse (0.1 ms, 0.1–1 mA) was delivered through a pair of thin-gauge (200 μm) tungsten wires (0.5 MΩ). Stimulation electrodes were connected to a stimulation isolation unit (World Precision Instruments, A360) that was controlled by a computer using pClamp 11.0 software (Molecular Devices). For the simultaneous activation of the OT and PT terminals, the electrical activation was paired with 400 ms of 50 Hz trains of blue light pulses (1 ms pulse duration). For the simultaneous activation of retinal terminals and PT terminals, the following light activation protocol was used: 500 ms of 20 Hz red light pulses (1 ms duration), followed by 5 s of no light, followed by 500 ms of 20 Hz blue light pulses (1 ms duration), followed by 3 s of no light, followed by 1.5 s of continuous red light stimulation with simultaneous 500 ms of 20 Hz blue light pulses (1 ms duration) during the last 500 ms of the continuous red light stimulation.
Morphological analysis of cells filled during electrophysiological recording
Following recording, slices were placed in a fixative solution of 4% paraformaldehyde in PB for at least 24 h. The sections were then rinsed in PB and incubated overnight in a 1:1,000 dilution of streptavidin-conjugated to Alexa Fluor-633 (Invitrogen) in PB containing 1% Triton X-100. The following day the slices were washed in PB, preincubated in 10% normal goat serum (NGS) in PB, and then incubated overnight in a 1:500 dilution (0.5 μg/ml) of a rabbit anti-DSred antibody (Clontech Laboratories, catalog #632496, RRID:AB_10015246) or a rabbit anti-GFP antibody (Millipore, catalog #AB3080, RRID:AB_91337) with 1% NGS. The following day the sections were rinsed in PB and incubated for 1 h in a 1:100 dilution of a goat-anti-rabbit antibody conjugated to Alexa Fluor-546 or Alexa Fluor-488 (Invitrogen). The sections were then rinsed in PB and mounted on slides to be imaged with a confocal microscope (Olympus FV1200BX61). Confocal images of labeled cells were categorized based on the following criteria: location of soma, orientation, and spread of dendritic fields or the presence or absence of GFP in the soma. Confocal and CCD images of each slice were aligned with outlines of the dLGN or vTRN and biocytin-filled vTRN cells, dLGN interneurons, and dLGN relay cells were plotted. For cells that were not recovered, their location was plotted using the CCD images of the patch pipette location.
Experimental design and statistical analyses
For analysis of the colocalization of VGAT- and GFP-labeled terminals, confocal images were deconvoluted using AutoQuant X3 software (RRID:SCR_002465). Deconvoluted images were then imported into Imaris (Oxford Instruments, RRID:SCR_007370), and surfaces were created based on fluorescent label of VGAT and GFP. A third colocalization color channel was created representing the double labeled VGAT and GFP terminals. Individual color channels were manually thresholded to create the colocalized surfaces which were then thresholded by size to label discrete terminals only. Surfaces labeling terminals from the colocalization channel were subtracted from the individual color channels labeled with VGAT and GFP to count the relative proportions.
For electron microscopic analysis of BDA, GFP, APEX2, and/or GABA-labeled terminals and their postsynaptic targets, ultrathin tissue sections were examined using an electron microscope and every labeled terminal involved in a synapse was imaged (n = 529 terminals). The pre- and postsynaptic profiles were characterized on the basis of size (measured using Image J, RRID: nif-000-30467) and the qualitative density of synaptic vesicles and overlying gold particles. A Kruskal–Wallis test with Dunn's multiple-comparisons test was used for statistical analyses of ultrastructural data and plotted as a histogram displaying the mean, standard error, and individual values using GraphPad Prism 10.4.
For analysis of the effects of PT photoactivation on the firing rate of relay, interneuron, or vTRN neurons, the percent reduction in firing rate during periods of light stimulation was compared with those during control periods utilizing a repeated-measures one-way ANOVA with Dunnett's multiple-comparisons test. For the analysis of the effects of PT photoactivation on responses evoked by photoactivation of terminals originating from V1 or retina, or by electrical stimulation of the OT, a paired t test was used to compare response amplitudes during activation of these inputs alone or during simultaneous activation of PT terminals. For analysis of the influence of PT terminal activation on spontaneous IPSCs, template matching was utilized to identify spontaneous inhibitory events during periods before, during, and after photoactivation of PT terminals, and the number of events during these periods was compared utilizing a repeated-measures one-way ANOVA with Sidak's multiple-comparisons test.
Results
GABAergic PT neurons innervate the dLGN and vTRN
The mouse visual thalamus is innervated by PT neurons located primarily in the nucleus of the optic tract (NOT), as demonstrated by injections of a retrogradely transported herpes simplex virus (HSV) that expresses Cre-recombinase in the dLGN of Ai9 mice (Fig. 1A). Labeled neurons are distributed primarily in the ipsilateral NOT, but a smaller population of neurons is labeled in the contralateral NOT. Subsequent injections of a Cre-dependent adeno-associated virus (AAV) centered in the NOT of GAD2-cre mice (Fig. 1B) labeled projections from the PT to the ipsilateral dLGN, ventral lateral geniculate nucleus (vLGN; Fig. 1C,D), superficial layers of the superior colliculus (SC, Fig. 1E), zona incerta (ZI, not shown), and vTRN (Fig. 2G–L). GABAergic PT axons are also labeled in the posterior commissure with sparser GABAergic PT projections distributed in the contralateral dLGN, SC, ZI, vLGN, and vTRN. Sparse GABAergic PT projections were also labeled bilaterally in the pulvinar, laterodorsal, and centromedial thalamic nuclei (data not shown).
GABAergic projections of the mouse pretectum (PT). A, PT cells labeled following injections of a retrogradely transported herpes simplex virus that expresses Cre-recombinase in the dLGN of an Ai9 mouse. A Cre-dependent AAV injection in the PT of a GAD2-cre mouse (B) labels axons and terminals in the dorsal lateral geniculate nucleus (dLGN) and ventral lateral geniculate nucleus (vLGN; C, rostral section, D, caudal section), and stratum griseum superficiale (SGS) of the superior colliculus (E). All sections are cut in the coronal plane. APN, anterior pretectal nucleus; NOT, nucleus of the optic tract; OPN, olivary pretectal nucleus; PC, posterior commissure; PUL, pulvinar nucleus; SGI, stratum griseum intermediale. Scale bars: A, 100 μm and also applies to B. Scale bars: C, 100 μm and also applies to D and E.
Terminals in the vTRN are labeled in the GAD67-GFP line and mimic GABAergic PT projections. In a GAD67-GFP mouse, the pulvinar nucleus (PUL) was injected with cholera toxin subunit B conjugated to Alexa Fluor 546 (A, C, E, red, CTB) to distinguish the visual sector of the thalamic reticular nucleus (vTRN), and the sections were stained with an antibody against parvalbumin to distinguish the borders of the TRN (A, C, E, blue, PV). The GFP label alone is shown in panels B, D, and F (green). In the GAD67 mouse line, GFP-labeled neurons are found in the dLGN and vLGN and GFP-labeled terminals are found in the vTRN. In a GAD2-Cre mouse, the pretectum was injected with a virus to induce the expression of a red fluorescent protein in a Cre-dependent manner (G–L, pseudocolored magenta, PT), and the sections were stained with an antibody against parvalbumin to distinguish the TRN (G, I, K, green, PV). The virus label alone is shown in panels H, J, L (magenta, PT), which mimics the distribution of GFP-labeled terminals in the vTRN of GAD67-GFP mice (B, D). Sections are arranged from rostral (A, B, G, H) to caudal (E, F, K, L). LD, laterodorsal nucleus. Scale bars: A, 100 μm and applies to A–F. Scale bars: G, 100 μm and applies to G–L.
The GAD67-GFP line labels GABAergic PT neurons and their axon projections
In the GAD67-GFP line, GFP-labeled axons overlap the distribution of CTB transported from the dLGN and/or PUL, which delineates the extent of the vTRN (Fig. 2A–F). Thus, the pattern of GFP-labeled axons in the GAD67-GFP line provides a convenient means to identify the vTRN. The distribution of GFP-labeled axons in the vTRN in the GAD67-GFP line mimics the GABAergic PT projections to the vTRN revealed by virus injections in the PT of GAD2-cre mice (Fig. 2G–L), suggesting that these GFP-labeled terminals originate from the PT. In fact, the PT of GAD67-GFP mice contains a dense population of GFP-labeled neurons (Fig. 3A). To determine whether these cells do in fact project to the vTRN, we injected the PT with Cre-dependent on green (Cre-DOG) viruses followed by a Cre-dependent virus to induce the expression of the red fluorescent protein TdTomato. These injections resulted in the expression of TdTomato in GFP-labeled PT cells (Fig. 3B,C) and revealed that they project to the dLGN and vTRN (as well as the other PT targets labeled in GAD2-cre mice). In the vTRN (Fig. 3D–F), the axons labeled by these virus injections were distributed specifically within the regions innervated by axons labeled in the GAD67-GFP line (Fig. 3G–I).
In GAD67-GFP mice, GFP-labeled PT neurons project to the vTRN. A, GFP-labeled cells (green) in the PT of a GAD67-GFP mouse are primarily located in the nucleus of the optic tract (NOT). The section was stained with an antibody against parvalbumin (magenta, PV) which labels neurons in the anterior pretectal nucleus (APN). B, Injection site in the NOT of a GAD67-GFP mouse with Cre-dependent on green (Cre-DOG) viruses and a Cre-dependent virus to express mCherry (magenta). C, The injection site illustrated in panel B in a 1 μm optical section. All mCherry-labeled cells (magenta) also contain GFP (green). The virus injections illustrated in B and C resulted in the labeling of axons in the vTRN (D, rostral through F caudal). The axons are pseudocolored green (PT) and the sections are stained for parvalbumin (magenta, PV) to delineate the borders of the TRN. G–I, The distribution of GFP-labeled axons (green) and parvalbumin (magenta, PV) in the sections illustrated in panels D–E. Scale bars: A, 100 μm; B, 50 μm; C, 20 μm; D, 50 μm and also applies to E–I.
The PT provides the majority of GABAergic input to the vTRN
To determine the proportion of GABAergic synaptic terminals within the vTRN that originate from the PT, we stained tissue from GAD67-GFP mice with an antibody against the vesicular GABA transporter (vGAT), which is contained within GABAergic synaptic terminals. As illustrated in Figure 4A–C, in the GAD67-GFP line the majority of vGAT antibody-stained terminals also contained GFP (of 11,662 vGAT terminals examined in three sections, 77.15 ± 12.16% contained GFP). In contrast, when the same analysis was carried out in the GAD65-GFP mouse line in which the majority of TRN neurons contain GFP (Fig. 4D–F; 64.12 ± 6.32% of vTRN cells labeled with a NeuN antibody contained GFP, n = 570 GAD65 + NeuN cells, n = 319 NeuN only, six sections), no terminals labeled with the vGAT antibody also contained GFP (Fig. 4G–I). Collectively, these results indicate that the PT provides the majority of GABAergic input to the vTRN.
The PT provides the majority of the GABAergic input to the vTRN. Sections that contain GFP-labeled terminals in the vTRN of GAD67-GFP mice (A, green) were stained with an antibody against the vesicular GABA transporter (vGAT, B, magenta) and the majority of vGAT labeled terminals contained GFP (C, white, of 11,662 vGAT terminals examined in 3 sections, 77.15 ± 12.16% contained GFP). Sections that contain GFP-labeled cells in the vTRN of GAD65-GFP mice (D, green) were stained with an antibody against neuronal nuclei (NeuN, E, magenta) and the majority of NeuN-labeled cells contained GFP (F, white, of 889 NeuN-labeled cells examined in 6 sections, 64.12 ± 6.32% contained GFP). Sections that contain GFP-labeled cells in the vTRN of GAD65-GFP mice (G, green) were stained with an antibody against vGAT (H, magenta); no vGAT terminals contained GFP (I). Scale bars: A, 20 μm and also applies to B, C, and G–I. Scale bars: D, 20 μm and also applies to E and F.
In fact, when we simultaneously labeled both GABAergic projections from the PT and GABAergic projections from other sources (Asanuma, 1994; Herrera et al., 2015; Nakajima et al., 2019), we found that they innervated different sectors of the TRN. Injections in the PT (Fig. 5A) innervate the vTRN (Fig. 5C) while paired injections in the lateral hypothalamus (LH; Fig. 5B) label axons that travel in the external medullary lamina (Fig. 5C) to innervate the anterior TRN (Fig. 5D). Large injections centered in the globus pallidus (Fig. 5F), with spread into the adjacent TRN and striatum, also failed to label projections to the vTRN. These large injections labeled terminals in the substantia nigra, parafascicular nucleus (Fig. 5E), and ventral TRN (Fig. 5H), while paired injections in the PT (Fig. 5E) labeled projections to the ZI (Fig. 5G) and vTRN (Fig. 5G,H).
Extrinsic GABAergic input to the TRN is sector specific. The pretectum (PT, A) and lateral hypothalamus (LH, B) of a GAD2-Cre mouse were injected with Cre-dependent viruses to induce the expression of GFP (green) and TdTomato (red) respectively, and tissue sections were stained with an antibody against parvalbumin to distinguish the TRN (blue). GABAergic projections from the PT labeled terminals in the dorsal, visual, TRN (C, green) while GABAergic projections from the LH labeled axons in the external medullary lamina (EML, C, red) that traveled to innervate the anterior TRN (D, red). Dual injections in the PT (E) and external globus pallidus (GPe, with spread into the adjacent TRN, and caudate putamen, CPu, F) of a GAD2-Cre mouse were injected with Cre-dependent viruses to induce the expression GFP (green) and TdTomato (red), respectively, and tissue sections were stained with an antibody against parvalbumin to distinguish the TRN (blue). GABAergic projections from the PT labeled terminals in the dorsal, visual, TRN (G, H, green) and zona incerta (ZI, G green) while GABAergic projections from the GPe injection labeled axons in the parafascicular nucleus (PF, E, red), internal globus pallidus (GPi, G, red), and ventral TRN (H, red). All scale bars: 100 µm. Scale in C also applies to D. Scale in E also applies to F. Scale in G also applies to H.
Ultrastructure of the vTRN
To confirm these labeling patterns at the ultrastructural level, we stained tissue from GAD67-GFP and GAD65-GFP mice with an antibody against GFP (tagged with an electron dense diaminobenzidine, DAB, reaction) and embedded reacted tissue for electron microscopic examination. As described above, in the vTRN of GAD67-GFP mice, GFP is contained in boutons (Fig. 6A,B). Ultrastructural examination of GAD67-GFP-DAB–labeled profiles in the vTRN revealed that labeled profiles were axons or synaptic terminals (n = 79 presynaptic terminals; Fig. 6C). In contrast, within the vTRN of GAD65-GFP mice, GFP is contained within somata and dendrites (Fig. 6D,E). Ultrastructural examination of GAD65-GFP-DAB–labeled profiles in the vTRN revealed no profiles filled with vesicles. Instead, all GAD65-GFP-DAB labeling sampled was confined to somata and dendrites (Fig. 6F; n = 151 profiles).
Synaptic organization of GAD67-GFP, GAD65-GFP, GABAergic, and nonGABAergic profiles in the vTRN. In the GAD67-GFP mouse line, GFP is contained within terminals in the vTRN (A, B). A diaminobenzidine (DAB) reaction was used to reveal GFP in tissue from GAD67-GFP mice, and sections prepared for electron microscopy were additionally stained to reveal GABA using an antibody tagged with gold particles. All DAB-labeled profiles observed in the vTRN of GAD67-GFP mice were GABAergic presynaptic terminals (n = 79, example illustrated in C, synapse indicated by the arrow). In the GAD65-GFP mouse line, GFP is contained within cells and dendrites in the TRN (D, E). Tissue from GAD65-GFP mice was prepared for electron microscopy as described above. All DAB-labeled profiles observed in the vTRN of GAD65-GFP mice were postsynaptic GABAergic somata and dendrites (n = 151; F, nonGABAergic synaptic inputs to the GAD65-GFP dendrite indicated by arrows). Surrounding terminals (n = 1,001) could be identified as GABAergic (high density of gold particles, 40.56%) or nonGABAergic (low density of gold particles, 59.44%). GABAergic terminals in the vTRN contain a high density of synaptic vesicles (F profiles) and contact somata (G) and dendrites (H and J). NonGABAergic terminals the vTRN contain a high density (D, examples illustrated in H and I) or low density (L, examples illustrated in I and J) of synaptic vesicles. Scale bars: A, 100 µm and also applies to D. Scale bars: B, 10 µm and also applies to E. Scale bars: C, 600 nm and applies to F. Scale bars; J, 600 nm and also applies to G–I.
All tissue prepared for electron microscopy was additionally stained to reveal GABA using an antibody tagged with gold particles. As illustrated in Figure 6C,F, there was a high density of gold particles overlying all DAB-labeled profiles, confirming that the GFP labels GABAergic elements within the vTRN of GAD67-GFP and GAD65-GFP mice. Photomontages of the GAD65-GFP vTRN tissue were collected and GABAergic and nonGABAergic terminals were identified (based on the presence of synaptic vesicles and either a high or low density of overlying gold particles, respectively) and counted. A total 1,001 terminals were identified and 406 (40.56%) contained GABA.
Examination of GABAergic terminals involved in synapses (n = 99) revealed that the majority of GABAergic terminals in the vTRN contain densely packed vesicles and contact the dendrites and somata of TRN neurons with symmetrical synapses (Fig. 6G,H,J), corresponding to the F type terminals previously described (Pinault et al., 1997). We also observed nonGABAergic D-type (densely packed vesicles; Fig. 6H,I) and L-type (loosely distributed vesicles; Fig. 6I,J) terminals that contact TRN dendrites with asymmetric synapses, presumably corresponding to cortical and thalamic inputs, respectively (Ohara, 1988). Taken together, these results suggest that the PT provides over 30% of the total synaptic input to the vTRN (Fig. 8A).
Synaptic organization of PT terminals in the vTRN and dLGN
To more specifically investigate the synaptic organization of PT terminals in the vTRN and also examine the PT innervation of the dLGN, unilateral iontophoretic injections of biotinylated dextran amine (BDA) were placed in the PT, and labeled terminals in the ipsilateral vTRN and dLGN were examined in tissue stained for GABA with gold particles. To relate the sizes of PT terminals and their postsynaptic targets to other terminal types in the vTRN or dLGN, terminals labeled via virus expression in V1 L6 cortical neurons or retinal ganglion cells (data from Whyland et al., 2022) were used for comparison. BDA-labeled PT terminals were densely distributed in the vTRN and were labeled with a high density of overlying gold particles (Fig. 7D,E). We examined a total of 90 PT terminals engaged in synapses in the vTRN and measured the sizes of the presynaptic terminals (0.52 ± 0.27 μm2) and their postsynaptic targets (1.10 ± 1.56 μm2). These sizes were not found to be significantly different from the sizes of the GAD67-GFP-DAB–labeled terminals in the vTRN (0.63 ± 0.37 μm2) or their postsynaptic targets (1.09 ± 1.30 μm2). The sizes of PT and GAD67-GFP-DAB terminals were also not significantly different from the sizes of the overall population of GABAergic terminals in the vTRN (0.62 ± 0.38 μm2). In contrast, the sizes of PT, GFP, or GABAergic presynaptic terminals in the vTRN were found to be significantly larger than presynaptic terminals originating from L6 of V1 (0.29 ± 0.14 μm2; n = 111; Fig. 8B; Kruskal–Wallis test; p < 0.0001). Additionally, the sizes of the postsynaptic targets (1.88 ± 2.24 μm2) of GABAergic terminals in the vTRN were not found to be significantly different than those targeted by the PT or GAD67-GFP-DAB terminals (Fig. 8C; Kruskal–Wallis test; p > 0.05) but were found to be significantly larger than the postsynaptic targets of terminals originating from V1 L6 (Fig. 8C; Kruskal–Wallis test; p = 0.018). These results indicate that PT and V1 terminals target similar dendritic compartments on vTRN neurons, corroborating previous comparisons of V1 and GABAergic terminals in the TRN (Liu and Jones, 1999) and suggesting that PT terminals are well positioned to influence cortex to vTRN transmission.
Synaptic targets of tracer labeled PT terminals in vTRN and dLGN. A–E, Terminals labeled following iontophoretic biotinylated dextran amine injections in the PT were visualized with a diaminobenzidine reaction (electron dense label) and the tissue was subsequently stained with an antibody conjugated to gold particles. All PT terminals are GABAergic (high density of overlying gold particles). In the dLGN PT terminals contact (arrows) nonGABAergic (low density of gold particles) dendrites (A) and the GABAergic dendrites of interneurons that contain vesicles (B, C). Interneuron dendrites postsynaptic to PT terminals can occasionally be seen to contact nonGABAergic dendrites (C, arrow with asterisk). In the TRN, PT terminals contact (arrows) GABAergic dendrites (D, E). Scale bars: A and D, 600 nm; B, 400 nm and also applies to C. Scale bar: E, 800 nm.
Proportions and sizes of PT inputs to the vTRN and dLGN. A, The PT provides over 30% of the synaptic input to the vTRN. B, There is no significant difference in the sizes of presynaptic terminals in the vTRN labeled via BDA injections in the PT (n = 90), GFP antibody staining of terminals in the GAD67-GFP mouse line (n = 79), or GABA antibody staining of terminals in the vTRN (n = 99), but each group is significantly larger than L6 V1 terminals (n = 111). C, There is no significant difference in the sizes of vTRN postsynaptic profiles innervated by PT terminals, GAD67-GFP terminals, or GABAergic terminals. The postsynaptic dendrites of vTRN neurons innervated by GABAergic terminals were significantly larger than those targeted by terminals originating from L6 V1. D, PT input to dLGN interneurons is approximately eightfold higher than PT input to relay cells. E, PT terminals that innervate interneurons (n = 50) are significantly smaller than PT terminals that innervate relay cells (n = 100) and both are significantly smaller than retinal (RT) terminals that innervate relay cells or interneurons. F, The dendrites of interneurons innervated by PT or RT terminals are significantly smaller than the dendrites of relay cells innervated by PT or RT terminals. There is no significant difference in the sizes of relay cell or interneuron dendrites innervated by PT or RT terminals. All comparisons carried out using Kruskal–Wallis test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Shown are mean and standard error of the mean.
In the dLGN, the majority of the BDA-labeled terminals (Fig. 7A–C) were GABAergic (identified by a high density of overlying gold particles). An additional population of BDA-labeled terminals contained pale mitochondria and the overlying density of gold particles was low. These were identified as retinal terminals (presumably labeled via axon collaterals that innervate the PT and dLGN; Dankowski and Bickford, 2003) and were not analyzed further. The postsynaptic targets of the GABAergic PT terminals were identified as both nonGABAergic relay cell dendrites (Fig. 7A), identified by a low density of overlying gold particles (n = 100), and interneurons (Fig. 7B,C), identified by a high density of overlying gold particles (n = 50). Since interneurons make up only 6% of the neurons in the dLGN (Evangelio et al., 2018), the PT input to each dLGN interneuron is approximately eightfold higher than the PT input to each dLGN relay cell (Fig. 8D).
The sizes of PT terminals that innervate interneurons (0.48 ± 0.21 μm2) were significantly smaller than the sizes of PT terminals that innervate relay cells (0.81 ± 0.44 μm2) and the sizes of interneuron dendrites postsynaptic to PT terminals (0.73 ± 0.83 μm2) were significantly smaller than relay cell dendrites postsynaptic to PT terminals (1.33 ± 1.27 μm2; Fig. 8E,F; Kruskal–Wallis test; *p = 0.045, ***p = 0.0006). Furthermore, the sizes of PT terminals innervating relay cells or interneurons were found to be significantly smaller than retinal (RT) terminals innervating relay neurons (2.86 ± 1.96 μm2) and interneurons (3.05 ± 1.96 μm2), respectively (Fig. 8E; Kruskal–Wallis test; PT vs RT Relay ****p < 0.0001; PT vs RT INT ****p < 0.0001). However, the sizes of relay cell or interneuron dendrites postsynaptic to PT or RT terminals were not significantly different, indicating these terminals target similar regions of the dendritic arbors of these two cell types (Fig. 8F; Kruskal–Wallis test; p > 0.05). Therefore, PT terminals are well positioned to influence retinogeniculate transmission. Finally, the majority (n = 44/50 or 88%) of the interneuron dendrites postsynaptic to PT terminals were observed to contain synaptic vesicles (Fig. 7B,C) and occasionally were observed to synapse on relay cell dendrites (Fig. 7C). Thus, PT synapses on interneurons target dendritic terminals.
Experimental paradigms for in vitro optogenetic activation of PT terminals
To determine how activation of PT terminals affects dLGN and vTRN cells, we injected Cre-dependent AAVs into the PT of GAD2-cre mice or Cre-independent viruses in the PT of GAD67-GFP mice to express channelrhodopsin (ChR2) or the ChR2 variant Chief and fluorescent proteins in PT axons and terminals. We then prepared coronal slices that contained the vTRN or dLGN, where we obtained current-clamp (K+-based internal solution) or voltage-clamp (Cs++-based internal solution) recordings from dLGN or vTRN neurons and activated PT terminals using blue light (schematically illustrated in Fig. 9A). In a second set of experiments, we paired Cre-dependent AAV injections in the PT to express the ChR2 variant Chronos and Cre-independent virus injections in V1 to express the ChR2 variant Chrimson in GAD2-cre mice (Fig. 9B) to simultaneously activate PT terminals with blue light and V1 terminals with red light. In a third set of experiments, we injected Cre-dependent AAVs into the PT of GAD2-cre mice to express ChR2 or Chief in PT axons and terminals and the optic tract was electrically activated (Fig. 9C). In the final set of experiments, we paired Cre-dependent AAV injections in the PT to express ChR2 and Cre-independent virus injections in the retina to express the Chrimson in GAD2-cre mice (Fig. 9D) to simultaneously activate PT terminals with blue light and retinal terminals with red or blue light. We included biocytin in our pipettes to subsequently examine the morphology of the recorded neurons and plot their location on templates of the vTRN (Fig. 9E), rostral dLGN (Fig. 9F), and middle dLGN (Fig. 9G). GAD67-GFP mice were used to target recordings to dLGN interneurons which contain GFP in this line (Fig. 9H); these neurons had relatively small somata and exhibited thin widespread dendritic arbors with appendages (Fig. 9I). In GAD67-GFP or GAD2-cre mice, recorded dLGN relay cells were identified as neurons with large somata and multipolar dendrites (Fig. 9J); recorded vTRN neurons displayed elongated dendritic fields that followed the contours of the vTRN (Fig. 9K).
Experimental paradigms for in vitro optogenetic activation of PT terminals. A–D, Schematics of optogenetic experiments carried out in slices of the dLGN and/or vTRN. In one set of experiments, we injected Cre-dependent AAVs into the PT of GAD2-cre mice or Cre-independent viruses in the PT of GAD67-GFP mice to express ChR2 or Chief in PT axons and terminals. We then prepared coronal slices, where we obtained recordings from dLGN or vTRN neurons and activated PT terminals using blue light (A). In a second set of experiments, we paired Cre-dependent AAV injections in the PT to express Chronos and Cre-independent virus injections in V1 to express Chrimson in GAD2-cre mice. Recordings were obtained from vTRN neurons while simultaneously activating PT and V1 terminals with blue and/or red light (B). In a third set of experiments, we injected Cre-dependent AAVs into the PT of GAD2-cre to express ChR2 or Chief in PT axons and terminals. While recording from dLGN neurons, the optic tract was electrically activated, and PT terminals were activated with blue light (C). In the final set of experiments, we paired Cre-dependent AAV injections in the PT to express ChR2 and Cre-independent virus injections in the retina to express the Chrimson in GAD2-cre mice. While recording from the dLGN, retina and PT terminals were activated with blue and/or red light pulses. For all recordings, biocytin was included in the pipettes to subsequently examine the morphology of the recorded neurons and plot their location on templates of the vTRN (E, legend indicates responses of plotted neurons), rostral dLGN (F, legend indicates responses of plotted dLGN neurons), and caudal dLGN (G). H, Interneurons in the dLGN were targeted for recording in GAD67-GFP mice which contain GFP in interneurons (green, biocytin-filled neuron in magenta). I, The interneuron illustrated in H is shown to illustrate the morphology of interneurons, which exhibit thin widespread dendritic arbors with appendages. J, Recorded dLGN relay cells were identified as neurons with large somata and multipolar dendrites. Shown are three relay cells (green) surrounded by PT terminals (magenta). K, Recorded vTRN neurons displayed elongated dendritic fields that followed the contours of the vTRN. Shown are two vTRN cells (green) surrounded by V1 terminals (magenta). Scale bars: H, 50 µm and also applies to I–K.
In vitro optogenetic activation of PT terminals inhibits interneurons, dLGN relay cells, and vTRN neurons via GABA release and reduces their firing rates
In interneurons, dLGN relay cells, and vTRN neurons held at 0 mV, activation of PT terminals with blue light pulses (1 ms pulses at 20 Hz), resulted in inhibitory postsynaptic currents (IPSCs; Fig. 10A–C) that could be blocked by the application of the GABAA receptor antagonist SR95531 (Fig. 10D–F; n = 5 vTRN neurons, n = 4 relay cells, n = 3 interneurons). To further examine the effect of PT input on these three cell types, we recorded in current-clamp mode to examine how PT photoactivation affects spiking activity. Neurons were given a 1.5 s square wave depolarizing current pulse of sufficient strength to evoke a steady train of spike firing >5 Hz (Fig. 10G–I). Changes in spike firing were then calculated (as in Campbell et al., 2020, 2024) by comparing equivalent periods (0.5 s) of activity in the presence or absence of blue light stimulation (2, 5, 10, 20, and 50 Hz; Fig. 10J–L, 50 Hz). To minimize the effects of frequency accommodation or burst firing, measures of spike firing were obtained 0.5 s after the onset of the 1.5 s current pulse.
Optogenetic activation of PT terminals evokes IPSCs in dLGN interneurons, relay cells, and vTRN neurons and reduces their firing frequency. During voltage-clamp recordings, optogenetic activation of PT terminals evokes inhibitory postsynaptic currents (IPSCs) in dLGN interneurons (A), dLGN relay cells (B), and vTRN neurons (C) that are blocked by the bath application of the GABAA receptor antagonist SR95531 (D–F). All cells held at 0 mV and responses to 20 Hz photoactivation of PT terminals (indicated by tick marks below traces) are shown. During current-clamp recordings, current steps were injected in dLGN interneurons (G), dLGN relay cells (H), and vTRN neurons (I) to induce action potential firing and PT terminals were activated with 2–50 Hz trains of blue light pulses during the current step (J–K show the firing of the cells in G–I during 50 Hz trains, indicated by the tick marks). The firing frequency during PT stimulus trains was calculated as a percentage of the firing frequency without PT input (M–O, each graph shows the mean and standard deviation). Most stimulus trains significantly reduced the firing frequency of dLGN interneurons (n = 21), dLGN relay cells (n = 17), and vTRN neurons (n = 17) when compared with control (repeated-measures one-way ANOVA with Dunnett's multiple-comparisons, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). P–R, The reduction in spiking during 20 Hz PT photoactivation (calculated as a percent of control) did not vary significantly as a function of initial control firing frequency for each cell type (simple linear regression, INT R2 = 0.018, Relay R2 = 0.026, vTRN R2 = 0.022).
As shown in Figure 10M–O, PT stimulus trains ≥2 Hz significantly reduced the firing frequency of dLGN interneurons (n = 21) and PT stimulus trains ≥5 Hz significantly reduced the firing frequency of vTRN neurons (n = 17) when compared with control (repeated-measures one-way ANOVA with Dunnett's multiple-comparisons, INT: p = 0.0153, 0.0016, 0.0005, <0.0001, <0.0001; Relay: p = 0.0410, 0.0066, 0.6121, 0.0004, 0.0251; vTRN: p = 0.0618, 0.0363, 0.0008, 0.0005, 0.0007). Most stimulus trains also significantly reduced the firing frequency of dLGN relay cells compared with control (n = 17, repeated-measures one-way ANOVA with Dunnett's multiple comparisons), but this was more variable, likely due a complex combination of direct PT inhibition and indirect disinhibition (via the inhibition of dLGN interneurons). Finally, for all three cell types, the reduction in firing rates during 20 Hz PT photoactivation (Fig. 10P–R) was not correlated with the initial control firing rates (simple linear regression, INT R2 = 0.018, Relay R2 = 0.026, vTRN R2 = 0.022).
PT input reduces vTRN responses to V1 input
Next, we examined the effect of PT terminal activation on vTRN responses to excitatory input. For these experiments we paired Cre-dependent virus injections in the PT of GAD2-cre mice to express the ChR2 variant Chronos with Cre-independent virus injections in V1 to express the red light activated ChR2 variant Chrimson (Fig. 9B). As previously demonstrated, Chronos is activated by low levels of blue light which do not activate Chrimson, and Chrimson is activated by red light which does not activate Chronos (Klapoetke et al., 2014). Therefore, using this dual opsin approach, we could assess the impact of PT input on vTRN neuron responses to V1 input. In slices maintained in vitro, vTRN neurons were patched in current-clamp mode. Activation of Chrimson-expressing V1 terminals with single pulses of red light (228.7 mW/mm2) resulted in large amplitude excitatory postsynaptic potentials (EPSPs) in vTRN neurons. In control experiments (with V1 injections only), the amplitudes of V1 EPSPs were not changed in the presence of 50 Hz trains (400 ms, 1 ms pulse duration) of low level (10 mW/mm2) blue light (Fig. 11A). However, using these same parameters in dual injection experiments, the amplitudes of V1 EPSPs evoked by single red light pulses were reduced when Chronos-expressing PT terminals were simultaneously activated with the train of blue light pulses (Fig. 11B; 24 of 29 recorded neurons showed a reduction of 6.4–75.2%). This effect was found to be significant (Fig. 11C; n = 29, paired t test, ***p = 0.0005) indicating that the PT can decrease vTRN responses to excitatory inputs.
PT projections suppress vTRN cell responses to V1 input and enhance dLGN relay responses to optic tract stimulation. A, During current-clamp recordings, photoactivation of Chrimson-expressing V1 terminals with single pulses of red light (red tick mark) results in large amplitude excitatory postsynaptic potentials (EPSPs) in vTRN neurons that are not changed in the presence of 50 Hz trains of low level blue light pulses (blue tick marks). An EPSP resulting from red light activation of V1 input (black) is overlapped with the same red light activated V1 EPSP in the presence of low level blue light pulses (blue). B, In the presence of Chronos-expressing PT terminals, the same low level 50 Hz blue light trains reduced the amplitude of V1 EPSPs. An EPSP resulting from red light activation of V1 input (black) is overlapped with the same red light activated V1 EPSP which was reduced by the blue light activation of PT terminals (blue). This effect was found to be significant (C, n = 29, paired t test, ***p = 0.0005). D, During current-clamp recordings, electrical stimulation of the optic tract (OT, black arrow) evoked EPSPs in dLGN relay cells that are unaffected by 50 Hz blue light trains (blue tick marks). An EPSP resulting from OT stimulation (black) is overlapped the same OT activated EPSP in the presence of low level blue light pulses (blue). E, When ChR2 or Chief-expressing PT terminals are present, the amplitude of most OT-evoked EPSPs are increased during 50 Hz blue light trains. An EPSP resulting from OT stimulation (black) is overlapped with the same OT activated EPSP which was increased by the blue light activation of PT terminals (blue). This effect was found to be significant (F, n = 16, paired t test, *p = 0.0394).
PT input enhances dLGN relay cell responses to retinal input
To determine how PT input impacts retinogeniculate transmission, we injected Cre-dependent AAVs into the PT of GAD2-cre mice to express ChR2 or Chief in PT axons and terminals and the optic tract (OT) was electrically activated (Fig. 9C). The amplitude of OT-evoked EPSPs was recorded in current-clamp mode. Then the amplitude of OT-evoked EPSPs was measured when ChR2 or Chief-expressing PT terminals were simultaneously activated by 50 Hz light trains. In control experiments (with no PT injections), the amplitudes of OT EPSPs were not changed in the presence of 50 Hz trains blue light pulses (Fig. 11D). However, in the presence of ChR2 or Chief-expressing PT terminals, photoactivation of PT input increased the amplitude of OT-evoked EPSPs in most dLGN relay cells (Fig. 11E; 12 of 16 relay cells showed a 1.9–69.6% enhancement of OT-evoked EPSPs), presumably via disinhibition. This effect was found to be significant (Fig. 11F; n = 16, paired t test, *p = 0.0394).
PT input disinhibits dLGN relay cells
In the preparation of slices for in vitro recordings, all extrinsic inputs to the dLGN and vTRN are cut which allowed us to isolate the effects of PT photoactivation. However, intrinsic circuits involving interneurons are preserved in slices of the dLGN. To more directly test whether PT input disinhibits dLGN relay cells, we used a second dual opsin approach. We injected a Cre-independent AAV in the eyes of GAD2-cre mice to induce the expression of Chrimson in ganglion cells and their axons and terminals. We subsequently injected a Cre-dependent AAV bilaterally in the PT to induce the expression of ChR2 in PT axons and terminals (schematically indicated in Fig. 9D). We then prepared slices of the dLGN and obtained voltage-clamped recordings from relay cells held at −60 or 0 mV. For these experiments, we used both blue and red light at full power (107.2 and 228.7 mW/mm2, respectively).
In control experiments in which Chrimson was expressed in retinogeniculate (RT) terminals only, responses to both 20 Hz red and 20 Hz blue light pulses could be recorded from dLGN relay cells (Fig. 12A). When cells were held at −60 mV, RT EPSCs were activated by both red and blue light pulses, and when cells were held at 0 mv, IPSCs (resulting from the retinal activation of interneurons) could be recorded (Fig. 12A1,2). However, RT responses to blue light could be blocked using an occlusion procedure. The occlusion procedure consisted of a continuous 2 s pulse of red light, which after an initial release of neurotransmitter from the RT terminals (Fig. 12A3) blocked any responses to simultaneous blue light pulses (Fig. 12A4). Conversely, in control experiments in which ChR2 was only expressed in PT terminals, dLGN relay cells did not respond to red light pulses (Fig. 12B1) but did respond during blue light pulses, even in the presence of the long red light pulse (Fig. 12B2,4). In this case, no responses were recorded when cells were held at −60 mV while IPSCs resulting from the direct PT innervation of relay cells were recorded when cells were held at 0 mV.
PT projections inhibit retina-evoked interneuron input to dLGN relay cells. A, While performing voltage-clamp recordings from dLGN relay cells at −60 mV, activation of Chrimson-expressing retinal (RT) terminals results in excitatory postsynaptic currents (EPSCs). When the same cell is recorded at 0 mV, activation of Chrimson-expressing RT terminals results in inhibitory postsynaptic currents IPSCs via interneuron activation. RT terminals are activated by either 20 Hz red or blue light trains (red and blue ticks, A1, A2). A continuous 2 s pulse of red light blocks retinogeniculate responses (A3) so no response is recorded during a simultaneous 20 Hz blue light train (A4). B, ChR2-expressing PT terminals are not activated by 20 Hz red light trains (B1) but are activated by 20 Hz blue light trains (B2). A 2 s red light pulse (B3) does not affect ChR2-expressing PT terminals so that they can be activated by simultaneous 20 Hz blue light pulses (B4). At 0 mV, PT activation results in IPSCs in relay cells and no response is recorded at −60 mV. C, D, Voltage-clamp recordings from dLGN relay cells in the presence of both Chrimson-expressing RT terminals and ChR2-expressing PT terminals at 0 and −60 mV. RT terminals are activated by red light (C1, D1), both RT and PT terminals are activated by blue light (C2, D2). A continuous 2 s pulse of red light blocks RT responses (C3, D3). During the 2 s occlusion of RT responses, PT terminals are activated with 20 Hz trains of blue light (C4, D4). The amplitudes of RT evoked IPSCs recorded in dLGN relay cells (E) were reduced when PT terminals were simultaneously activated (F). This effect was found to be significant (G, n = 9, paired t test, *p = 0.0189). For relay cells that received direct PT input, blue light trains during the 2 s red light occlusion, evoked IPSCs (H). For relay cells that did not receive direct PT input, spontaneous IPSCs were reduced during photoactivation of PT terminals (I). This effect was found to be significant (J, n = 5, two-way repeated-measures ANOVA with Sidak's multiple-comparison test, pre vs PT stim *p = 0.0384, post vs PT stim **p = 0.0056, pre vs post p = 0.5040).
In dual opsin experiments in which Chrimson was expressed in RT terminals and ChR2 in PT terminals (Fig. 12C,D), we could determine the impact of PT input on interneuron inhibition of relay cells by comparing the amplitude of IPSCs evoked by red (Fig. 12C1,D1,E) or blue light pulses (Fig. 12C2,D2,F). The occlusion procedure then allowed us to examine responses of the same cells to activation of PT terminals alone (Fig. 12C4,H, direct response; Fig. 12D4,I, indirect response). This paradigm revealed that the amplitudes of IPSCs recorded in dLGN relay cells (resulting from the photoactivation of RT terminals, Fig. 12E) were reduced (15.35–66.42%) when PT terminals were simultaneously activated (Fig. 12F), and this effect was found to be significant (Fig. 12G; n = 10, paired t test, *p = 0.0124, for this comparison the IPSC amplitudes evoked by trains of 10 light pulses were averaged in each condition). Moreover, this effect was evident in cells that received no direct PT input (Fig. 12D4,I) as well as cells that received direct PT input (Fig. 12C4,H). Finally, for relay cells that did not receive direct PT input, there was a significant reduction in the number of spontaneous IPSCs during photoactivation of PT terminals when compared with before and after PT activation (Fig. 12J; n = 5, two-way repeated-measures ANOVA with Sidak's multiple-comparison test, pre vs PT stim *p = 0.0384, post vs PT stim **p = 0.0056, pre vs post p = 0.5040). Taken together, the results of this study provide evidence that activation of the PT disinhibits dLGN relay cells via two different circuit mechanisms: inhibition of vTRN cells and inhibition of dLGN interneurons.
Discussion
Our results support the following conclusions: (1) GABAergic PT neurons innervate the dLGN and vTRN; (2) GABAergic projections from the PT to the dLGN and vTRN are labeled in the GAD67-GFP mouse line; (3) PT provides over 30% of the total synaptic input to the vTRN and comprises over 75% of all inhibitory input to the vTRN; (4) PT and V1 terminals innervate similar dendritic compartments of vTRN cells; (5) PT and RT inputs innervate similar dendritic regions of relay cells and interneurons; (6) PT input to dLGN interneurons is approximately eightfold higher than input to relay cells; (7) PT input preferentially innervates the presynaptic dendritic appendages of interneurons; (8) optogenetic activation of PT terminals elicits IPSCs in vTRN cells, dLGN relay cells, and dLGN interneurons via the release of GABA; (9) optogenetic activation of PT input reduces dLGN relay cell firing but is most effective in reducing the firing rates of dLGN interneurons and vTRN neurons; (10) optogenetic activation of PT input reduces the responses of vTRN cells to simultaneous optogenetic activation of V1 input; (11) optogenetic activation of PT input increases the amplitude dLGN relay cell responses to electrical stimulation of the optic tract; and (12) optogenetic activation of PT terminals can reduce interneuron inhibition of dLGN relay cells. As summarized in Figure 13 and discussed below, these results, in combination with the results of previous studies, suggest that PT projections to the vTRN and dLGN can boost relevant visual signals following gaze shifts.
Summary of investigated circuits. Schematic diagram summarizes the GABAergic projections of the PT to the visual thalamus. The PT provides two routes for the disinhibition of dLGN relay cells which can boost the transfer of retinal signals to V1.
Visuomotor response properties of PT neurons
The PT contains a variety of cell types that can be distinguished by their location, projections, and/or responses to sensory signals and animal movement. For those PT neurons that respond to visual input, four main types have been identified. Neurons in the olivary pretectal nucleus (OPN) receive input from melanopsin-containing retinal ganglion cells and mediate the pupillary light reflex via projections to the Edinger–Westphal nucleus. These cells exhibit low firing rates in darkness and sustained responses to light; they are therefore referred to as “sustained-on” cells or luminance detectors (Young and Lund, 1994; Schmidt et al., 2011; Sun and May, 2014; Seabrook et al., 2017; Hayter and Brown, 2018; May and Warren, 2020). Another well-characterized PT cell type is the “retinal slip” neuron, which responds to slow horizontal movements of the retinal image and exhibits strong directional selectivity. “Retinal slip” neurons are nonGABAergic cells in the nucleus of the optic tract (NOT) that project to the ipsilateral inferior olive; these cells are essential for the control of the slow-phase eye movements of optokinetic nystagmus (Hoffmann and Schoppmann, 1981; Horn and Hoffmann, 1987; Liu et al., 2016).
Two other PT cell types, identified in the cat, have been proposed to mediate either the suppression of visual signals during rapid movements to shift gaze or the facilitation of visual signals from newly acquired targets following gaze shifts. PT “jerk” neurons respond with high-frequency bursts to fast stimulus displacements (“jerks”), during the execution of saccadic eye movements in the light but not in the dark, and after “on” or “off” visual stimulation (Schoppmann and Hoffmann, 1979; Schweigart and Hoffmann, 1992). PT “saccade” neurons are similar to “jerk” neurons, but they also respond to eye movements in the dark, a feature which suggests they receive both visual and extraretinal saccade-related signals (Schmidt, 1996). In the cat, based on antidromic stimulation and tracing studies, “jerk” neurons are nonGABAergic and project to the pulvinar nucleus, while “saccade” neurons are GABAergic and project to the dLGN (Cucchiaro et al., 1991; Wahle et al., 1994; Schmidt et al., 2001; Wang et al., 2002; Baldauf et al., 2005).
Saccadic suppression and postsaccadic facilitation of visual signals
In the cat, GABAergic PT “saccade” neurons contact GABAergic dLGN interneurons (Cucchiaro et al., 1993; Wang et al., 2002) and are therefore proposed to mediate a disinhibition of relay cells to allow them to be fully excitable after the execution of a saccade, i.e., after the center of gaze has changed and a completely new image is to be analyzed. Consistent with this proposed function, in the cat, the activity of most relay cells is facilitated after saccades (Fischer et al., 1996; Lee and Malpeli, 1998), and this facilitation is lost or reduced when the PT is inactivated via muscimol injections (although variations and cell type-specific effects have been identified; Funke and Eysel, 1995; Fischer et al., 1998).
Similarly, in the primate, both magnocellular and parvocellular dLGN neurons show weak suppression during saccades, followed by a strong enhancement after the completion of a saccade, when a new target location is acquired (Reppas et al., 2002). This is presumably also carried out via disinhibition of dLGN thalamocortical neurons. PT cells that project to the primate dLGN are GABAergic (Bickford et al., 2000), and ultrastructural studies of PT terminals in the primate dLGN have categorized these terminals as F1 profiles which are observed to synapse on the dendritic terminals of interneurons (Feig and Harting, 1994). However, PT terminals in the primate dLGN also contact conventional dendrites, and it is currently unknown whether these arise from relay cells or interneurons.
Possible functions of PT projections in the mouse
For the experiments carried out in frontal-eyed cats and primates described above, recordings were carried out in regions of the dLGN representing the binocular visual field (∼central 140°), and the conjugate saccadic eye movements tested did not extend outside this range. In the lateral-eyed mouse, the binocular visual field is much more restricted (central 50°; Priebe and McGee, 2014), and gaze shifts are primarily carried out via a combination of head and either conjugate or disconjugate eye movements (Meyer et al., 2020). Future in vivo studies of the visual and motor-related activity patterns of mouse PT neurons that project to the dLGN and vTRN are needed to further reveal the function of these projections. However, if mouse PT neurons fire during gaze shifts, an initial inhibition of dLGN relay cells during head movements might be followed by their disinhibition once the eyes are redirected to view attended targets within the binocular visual field.
Implications for the “searchlight” concept of TRN function
Our results indicate that the PT is the primary source of GABAergic input to the mouse vTRN, providing at least 30% of its total synaptic input; when this large projection was activated, it strongly reduced vTRN firing rates and responses to V1 input. Additionally, our results indicate that PT input is specific to the vTRN, with other sources of GABAergic input innervating the TRN in a sector-specific manner. A sector-specific organization of extrinsic GABAergic inputs could mediate the precise control of TRN activity proposed to create a “searchlight” of attention selection (Crick, 1984). Wimmer et al. (2015) demonstrated that mouse vTRN activity is modulated during a divided attention task; when attending to auditory signals, vTRN activity increased, and when attending to visual signals, vTRN activity decreased. GABAergic projections from the PT could suppress vTRN activity during such attention shifts, although it remains to be determined whether the PT is involved in attention selection or overt attention shifts (Li et al., 2021). Other sparser extrinsic and/or intrinsic GABAergic inputs to vTRN neurons may be involved in attention selection (Deleuze and Huguenard, 2006; Nakajima et al., 2019), while the PT modulation of the vTRN and dLGN may be specifically related to subsequent gaze shifts. In this case, the PT may mediate an “active searchlight of attention” that coordinates gaze shifts with the enhanced transfer of retinal information from attended visual targets to the cortex.
Footnotes
This work was supported by the National Eye Institute (R01EY031322, EY035523, and EY012716), the National Institute of Neurological Disorders and Stroke (R21NS104807), and the National Institute of General Medical Sciences (P20GM1034236). The authors thank Arkadiusz Slusarczyk for his assistance with histology and electron microscopy.
↵*J.B.W. and S.P.M. contributed equally to this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Martha Bickford at martha.bickford{at}louisville.edu.
