Abstract
The amygdala, hippocampus, and subgenual cortex area 25 (A25) are engaged in complex cognitive-emotional processes. Yet pathway interactions from hippocampus and A25 with postsynaptic sites in amygdala remain largely unknown. In rhesus monkeys of both sexes, we studied with neural tracers how pathways from A25 and hippocampus interface with excitatory and inhibitory microcircuits in amygdala at multiple scales. We found that both hippocampus and A25 innervate distinct as well as overlapping sites of the basolateral (BL) amygdalar nucleus. Unique hippocampal pathways heavily innervated the intrinsic paralaminar basolateral nucleus, which is associated with plasticity. In contrast, orbital A25 preferentially innervated another intrinsic network, the intercalated masses, an inhibitory reticulum that gates amygdalar autonomic output and inhibits fear-related behaviors. Finally, using high-resolution confocal and electron microscopy (EM), we found that among inhibitory postsynaptic targets in BL, both hippocampal and A25 pathways preferentially formed synapses with calretinin (CR) neurons, which are known for disinhibition and may enhance excitatory drive in the amygdala. Among other inhibitory postsynaptic sites, A25 pathways innervated the powerful parvalbumin (PV) neurons which may flexibly regulate the gain of neuronal assemblies in the BL that affect the internal state. In contrast, hippocampal pathways innervated calbindin (CB) inhibitory neurons, which modulate specific excitatory inputs for processing context and learning correct associations. Common and unique patterns of innervation in amygdala by hippocampus and A25 have implications for how complex cognitive and emotional processes may be selectively disrupted in psychiatric disorders.
SIGNIFICANCE STATEMENT The hippocampus, subgenual A25, and amygdala are associated with learning, memory, and emotions. We found that A25 is poised to affect diverse amygdalar processes, from emotional expression to fear learning by innervating the basal complex and the intrinsic intercalated masses. Hippocampal pathways uniquely interacted with another intrinsic amygdalar nucleus which is associated with plasticity, suggesting flexible processing of signals in context for learning. In the basolateral (BL) amygdala, which has a role in fear learning, both hippocampal and A25 interacted preferentially with disinhibitory neurons, suggesting a boost in excitation. The two pathways diverged in innervating other classes of inhibitory neurons, suggesting circuit specificities that could become perturbed in psychiatric diseases.
- emotional regulation
- fear conditioning
- inhibitory neurons
- neural circuits
- nonhuman primate
- ventromedial prefrontal cortex
Introduction
The amygdala, hippocampus, and subgenual cingulate area 25 (A25) of the medial prefrontal cortex (mPFC) are essential for processes associated with emotion, memory, and internal states, and disruption of these nodes can precipitate psychiatric disturbances in humans (for review, see Hiser and Koenigs, 2018; Gray et al., 2020; Kaul et al., 2021). Understanding the organization of these circuits in nonhuman primates remains an urgent goal, as novel and safe therapies for mood disorders have proven difficult to translate from rodent models, although remarkable strides have been made (Roberts, 2020).
In mice and rats, recent studies have highlighted a tight coupling between the hippocampus, amygdala, and the infralimbic cortex (IL) during fear-related processes (Lesting et al., 2011; Sierra-Mercado et al., 2011; Jin and Maren, 2015; W.B. Kim and Cho, 2017; Qi et al., 2018; Saha et al., 2018). Hippocampal signals to amygdala are thought to contribute contextual information to facilitate associative learning in amygdalar microcircuits (Maren and Fanselow, 1995; Seidenbecher et al., 2003; Knight et al., 2004; Smith et al., 2006; Wiltgen et al., 2006; Marschner et al., 2008; Orsini et al., 2011), while projections from rodent IL to amygdala are associated with fear inhibition and extinction (Giustino and Maren, 2015). The IL cortex is in some ways similar to primate A25, although important differences have emerged. In contrast to rodent IL, marmoset A25 is associated with driving activity in the amygdala and promoting autonomic responses to threats (Wallis et al., 2017; Alexander et al., 2019, 2020; Rudebeck and Izquierdo, 2022). In humans, A25 is associated with internal states and mood, and is abnormally hyperactive in depression (for review, see Hamani et al., 2011; Myers-Schulz and Koenigs, 2012). These crucial species differences suggest phylogenetic divergence in mPFC in primates. While aspects of the amygdalar nuclei show close homology across rodents and primates, the basolateral complex has expanded along with the cortex in primates (Johnston, 1923; Price, 1981; De Olmos, 1990; Teffer and Semendeferi, 2012; McDonald, 2020), leaving unresolved questions about intrinsic and extrinsic circuit mechanisms for fear processes (Duvarci and Pare, 2014).
A25 and the anterior hippocampus densely innervate the amygdala in macaque monkeys (Aggleton, 1986; Saunders and Rosene, 1988; McDonald, 1998; Ghashghaei et al., 2007; Fudge et al., 2012; Kelly et al., 2021), although it is unknown how these pathways interact with local excitatory and inhibitory neurons. In rats and mice, inhibitory neurons in the basolateral complex (BLcx, which includes basolateral, lateral, and basomedial nuclei) play an integral role in many aspects of fear learning paradigms, including acquisition, recall, and extinction, with some division of labor (Wolff et al., 2014; Krabbe et al., 2019; for review, see Krabbe et al., 2018; Ressler and Maren, 2019; Perumal and Sah, 2021). Failure to engage inhibition in the BLcx is associated with more excitable pyramidal neurons and anxiety-related behaviors (Hetzel and Rosenkranz, 2014; Rau et al., 2015), while engagement of inhibition in BL reduces fear and anxiety (Saha et al., 2017; Qi et al., 2018; Elorette et al., 2020; Roseboom et al., 2021).
Like in cortex, inhibitory neurons in macaque or human BLcx can be usefully classified by their expression of the calcium binding proteins (CBPs) parvalbumin (PV), calbindin (CB), and calretinin (CR; Pitkänen and Amaral, 1993a, b; Sorvari et al., 1995, 1996a, b; Pitkänen and Kemppainen, 2002; Muller et al., 2006; Mascagni et al., 2009; McDonald, 2021). Here, we mapped termination patterns of A25 and hippocampus across the amygdala, and quantitatively analyzed their synaptic interactions with excitatory neurons and putative inhibitory neurons labeled for the CBP in the amygdala. Our findings reveal amygdalar regions under common as well as distinct influence from these afferent structures. We also contrast the relationships of hippocampus and A25 with inhibitory neurons in BLcx, which suggest how distinct microcircuits could go awry in psychiatric diseases (Kenwood et al., 2022).
Materials and Methods
Experimental design
Figure 1 depicts our experimental approach to study how A25 and hippocampal axon terminations innervate the amygdala. We injected neural tracers in A25 and the hippocampus (Fig. 1A) and examined their anterogradely labeled terminations in amygdala (Fig. 1B). First we studied the patterns of these terminations across all nuclei in a series of sections throughout anterior to posterior amygdala (Fig. 1C,D). Then we used higher resolution methods to examine the synaptic targeting patterns of the two labeled pathways (Fig. 1E–G). Guided by previous studies (Barbas and De Olmos, 1990), we used comparison across multiple histochemical methods to parcellate the amygdala (e.g., Fig. 2).
Surgery, tracer injection, and perfusion
To study the pathways from A25 and hippocampus to amygdala, we conducted experiments on seven rhesus monkeys of both sexes (Macaca mulatta; n = 6 female; aged 3–5.5 years; Table 1) according to protocols approved by the Institutional Animal Care and Use Committee at Boston University School of Medicine, Harvard Medical School, New England Primate Research Center, in compliance with the Institute for Laboratory Animal Research (ILAR) Guide. Procedures were designed to minimize animal suffering. Each case received injections of multiple distinct tracers to address questions in this and other unrelated studies, thus minimizing the number of animals needed for research.
Cases used and analyses performed
For tracer injection (Table 1), animals were placed in a stereotaxic apparatus (1430 M, David Kopf Instruments) and underwent magnetic resonance imaging (after sedation, ketamine hydrochloride, 10–15 mg/kg, i.m., and anesthesia, propofol, loading dose 2.5–5 mg/kg, i.v., continuous infusion rate 0.25–0.4 mg/kg/min). MRI scans were registered to the stereotaxic space and used to plan stereotaxic coordinates for later injection of neural tracers. For surgery, animals were again sedated (as above) and placed under general anesthesia (isoflurane, to a surgical level). Animals were returned to the stereotaxic apparatus to match original MRI placement. We then performed surgery under sterile conditions with continuous monitoring of respiratory rate, oxygen saturation, heart rate, and temperature. A microdrive attachment was used to mount syringes for precise localization of planned injection coordinates. A small area of cortex above the designated injection sites was exposed, and microsyringes (Hamilton, 5 or 10 µl) loaded with tracer (Table 1, all Invitrogen) were mounted and guided toward the final injection site coordinates. For A25 injections, we used a dorsoventral approach, with the needle angled 10° laterally to avoid the superior sagittal sinus. For hippocampal injections, we used a straight dorsoventral needle trajectory. The needles passed through structures along the trajectory. To avoid tracer leakage during insertion, we loaded syringes with a small bubble of air after aspiration of the tracer. We injected 4–6 µl of each tracer at each injection site. Syringes were left at the site of injection for 5–10 min to allow for local diffusion of the tracer, and prevent backward suction of the tracer during syringe retraction. Animals were monitored postoperatively and given antibiotics and analgesics, as needed.
After a period of 18–20 d for tracer propagation, animals were sedated, given a lethal dose of anesthetic (sodium pentobarbital, to effect), and transcardially perfused (4% paraformaldehyde, 0.2% glutaraldehyde in 0.1 m PBS, pH 7.4; cases AY, BQ, BR, BS, BT, BU; 4% paraformaldehyde in 0.1 m PBS, pH 7.4, case BW). Brains were removed, photographed, and cryoprotected in ascending sucrose solutions (10–25% sucrose, 0.05% sodium azide in 0.1 m PB, pH 7.4). Arachnoid mater and microvessels above the pia were removed from the brain surface. The brain was then dried and deep-frozen in −80°C isopentane, and later sectioned on a freezing microtome (AO Scientific Instruments/Reichert Technologies) at 50 µm into 10 matched series. The series were placed in antifreeze (30% ethylene glycol, 30% glycerol, 0.05% sodium azide in 0.01 m PB, pH 7.4) for long-term storage until processing.
Histologic procedures for parcellating the amygdala
We used matched sections with Nissl staining, acetylcholinesterase (AChE), or wet-mounted tissue (e.g., Fig. 2) to parcellate the amygdala using classical criteria (Barbas and De Olmos, 1990; De Olmos, 1990). Wet-mounted tissue was pulled directly from long-term storage solution for quick photography in buffer on an unsubbed slide, which reveals the white matter in nucleated structures such as the amygdala, as demonstrated previously (Timbie and Barbas, 2015). We have previously published procedures used for Nissl (García-Cabezas et al., 2016) and AChE (Wang and Barbas, 2018).
Labeling procedures for brightfield, confocal, and electron microscopy
We used multiple labeling immunohistochemical techniques to visualize neural tracers and markers of inhibitory neurons at multiple scales. For brightfield microscopy axon tracing and stereological analysis, we used single immunoperoxidase labeling to visualize the tracers using diaminobenzidine (DAB). For fluorescence confocal microscopy (CM) appositional and bouton size analysis, we used multiple labeling via secondary antibodies conjugated with distinct fluorophores. For electron microscopy (EM) synaptic analysis, we used multiple labeling via secondary antibodies conjugated with gold, or with biotin for DAB and tetramethylbenzidine (TMB).
Procedures through the incubation of primary antibodies are as follows, while procedures after the primary incubation differ by visualization paradigm and are described sequentially below. First we rinsed free-floating sections of cryoprotectant using PB (0.1 m, pH 7.4). Antigen retrieval (not used for EM) was performed in sodium citrate solution (10 mm, pH 8.5, Sigma-Aldrich) using a 70–80°C water bath. Sections were rinsed and incubated in 0.05 m glycine (4°C, 1 h, Sigma-Aldrich) to bind free aldehydes. After rinsing, sections were incubated in 0.3% H2O2 in 0.1 m PB (30 min) to quench endogenous peroxidases (not used for fluorescence microscopy). In cases with a BDA injection not under study, we used avidin-biotin blocking solution (AB blocking, Vector, catalog #SP-2001, RRID:AB_2336231) to block the biotin conjugate on the tracer (not necessary for fluorescence microscopy). Sections were then rinsed and blocked in preblocking solution, which contained 5% bovine serum albumin (BSA; Sigma-Aldrich), 10% normal serum of the secondary antibody host animal (Sigma-Aldrich, donkey or goat), 0.2% BSA-c (Aurion), Triton X-100 (0.2% Sigma-Aldrich, for brightfield and confocal microscopy) or reduced Triton X-100 (0.025%, Roche Applied Science; for EM), and cold-water fish gelatin (0.1%, Aurion; for EM) for stabilization of ultrastructure. Sections were incubated with primary antibodies (Table 2), for 1–3 d, rinsed, then incubated in secondary antibodies (Table 2) for 3 h to 1 d. For control sections, we omitted the primary antibodies. Incubation was enhanced by microwaving (2–6 min at 150 W, 4°C, PELCO BioWave with SteadyTemp thermocooler, Ted Pella).
Antibodies used
To visualize tracers for brightfield microscopy axon tracing and stereological analysis (Table 1), we used a biotin-conjugated secondary antibody incubation. We then rinsed sections and incubated them in avidin-biotin horseradish complex (ABC, 1:100 in 0.1 m PB, AB-HRP, catalog #PK-6100, Vector, RRID: AB_2336827), followed by processing for DAB for 1–3 min (DAB kit, catalog #SK-4100, Vector, RRID: AB_2336382). Sections were mounted on gelatin-coated glass slides and dried. We counterstained some sections for Nissl (thionin stain) as previously described (García-Cabezas et al., 2016). Brightfield microscopy sections were coverslipped with Entellan (Sigma-Aldrich).
To visualize the tracer and PV, CB, or CR for fluorescence CM appositional and bouton size analysis (Table 1), we used secondary antibodies conjugated with fluorophores. After incubation, the sections were mounted, dried, and coverslipped with Prolong Gold Antifade (catalog #36930, Invitrogen) or FluoroSave (Calbiochem, Fisher Scientific catalog #345789) and edges were hardened using fast-drying clear nail polish (Sally Hansen), which helps prevent the infiltration of air bubbles into the cover-slipping medium.
To visualize structures for EM synaptic analysis (Table 1), we used double or triple labeling for tracer and PV, CB, and/or CR. The tracer was always visualized with DAB. For double labeling, the primary antibody for the next marker was paired with a gold-conjugated secondary antibody. For triple-labeling, the third marker was visualized with TMB. DAB forms a dark precipitate that fills the cytoplasm of neuronal elements containing tracer. Silver-enhanced gold particles appear as small quanta that sometimes cluster, and TMB forms black stripes or crystalline appearing structures. After primary antibody incubation (Table 2), we rinsed sections and incubated them in biotin-conjugated donkey anti-rabbit (for the tracer-DAB; Table 2), and a gold-conjugated secondary for the second marker. The sections were washed and underwent a rapid reduced concentration postfixation (3% glutaraldehyde, 1% paraformaldehyde in 0.1 m PB with 2-min microwave at 150 W, 4°C), followed by a glycine wash and rinse. We then silver-enhanced the gold secondary antibodies starting with an enhancement conditioning solution wash (ECS, 1:10, 10 min, Enhancement Conditioning Solution 10_500.055, Electron Microscopy Sciences, catalog #25830), silver enhancement (90 min, R-GENT SE-EM kit 500.033, Electron Microscopy Sciences, catalog #255213), ECS again, and 0.1 m PB rinses. Then the tissue underwent ABC and DAB, as described above, to visualize the tracer. If the tissue was being triple-labeled, sections were rinsed in 0.1 m PB, then underwent another 0.3% H2O2 to quench remaining peroxidases, and AB-blocking again to block remaining HRP-binding sites. We then incubated the tissue with the third secondary antibody, which had a biotin conjugate to visualize TMB. After incubation, sections underwent rinses, then TMB staining with stabilization using a DAB cobalt chloride solution, as previously described (Medalla et al., 2007).
After EM-immunolabeling, sections then require additional staining and fixation. First, we performed postfixation in the microwave (6% glutaraldehyde, 2% paraformaldehyde in 0.1 m PB, 150W, 15°C) until sample temperatures reached 30–35°C, and then left the sections in the fixative to come to room temperature for 30 min, followed by 0.1 m PB rinses. We conducted EM processing using either a routine protocol for transmission EM (A25, case BQ, BW; hippocampus, case BS) or a modified protocol optimized for high throughput block-face imaging in a scanning EM (hippocampus, case BQ). We have described routine and modified protocols for EM methods for ultrathin sectioning previously (Joyce et al., 2020; Wang et al., 2021).
Pathway mapping
In the ipsilateral hemisphere of each injection site, we immunohistochemically labeled for tracer in series of sections throughout the amygdala (one in 20 sections at 50 µm, or one section per 1 mm). These series of sections were then used for qualitative mapping of tracer-labeled terminations via manual tracing of individual axons and stereological quantification of axon terminations.
In two cases with hippocampal injections and three cases with A25 injections (hipp: BQ, BT; A25: BR, BS, BU; Table 1), we selected three representative sections containing amygdala for individual axon tracing. Axon tracing was performed using a microscope (Olympus BX60) coupled to a CCD camera and semiautomated commercial system (Stereo Investigator 10, MBF Bioscience, Vermont RRID:SCR_002526). We performed axon tracing using a meander-scan function at 1000× with oil immersion to exhaustively capture tracer-labeled bouton-bearing axons within the amygdala. In densely labeled ventral BM, BLpc, and PLBL of one hippocampal case, we used a qualitative assessment of density to apply a repeated modular axonal innervation pattern to multiple fields of view, and then adjusted mismatched terminations as needed. For all other cases, we traced labeled axons with boutons in their entirety, and avoided axons that did not have boutons that could be passing through. In order to fully characterize the termination patterns from sectors of A25, we added single sections from the midlevel amygdala in two additional A25 injection sites (cases BQ, BW).
To quantify the density of axon terminations in amygdalar nuclei for three hippocampus and three A25 cases (hipp: BQ, BS, BT; A25: BR, BS, BU; Table 1), we used the tracer labeled series (one in 20 sections) to perform stereology on axon terminations, or boutons (five sections/injection site). Stereological sampling was performed using commercially available software (Stereo Investigator 10, MBF Bioscience, RRID:SCR_002526), which uses the optical fractionator method to apply systematic random sampling at regular intervals to extrapolate a volumetric count of a structure of interest (here, tracer-labeled axon terminations) in each region of interest (here, amygdala nuclei). We used sampling parameters designed to minimize the Gundersen coefficient of error to <10%, as recommended (Gundersen, 1986). Hippocampal cases were sampled using a 0.5- to 1-µm guard zone, 3-µm disector height, a counting frame of 70 × 60 or 60 × 60 µm, and a sampling grid of 200 × 200 to 900 × 900 µm. A25 cases were sampled using a 2-µm guard zone, 10-µm disector height, a counting frame of 50 × 50 to 75 × 75 µm counting frame, and a sampling grid of 250 × 250 to 300 × 300 µm. We computed the volumetric density by dividing the extrapolated number of terminations by the estimated volume of the nucleus, which was then was averaged across pathways and expressed with standard error of the mean. For the A25 injections, a vector graphics map of each sampled amygdala section with grids of axon termination counts was exported and overlaid with axon tracing maps to fully illustrate the innervation patterns of A25 in the amygdala.
Appositional analysis
We imaged sections processed for CM appositional and bouton size analysis using a laser-scanning confocal microscope (Axio Observer Z1, LSM 880, Zeiss) at 630× (Zen 2.1 package, RRID:SCR_013672). Green fluorophores were imaged using a 488 nm argon ion laser, red fluorophores with a 568 DPSS 561–10 laser, blue fluorophores with a Diode 405–30 laser, and far-red fluorophores were imaged with a 633-nm helium neon laser (all Zeiss). We acquired stacks of optical sections (0.31- or 0.33-µm step size). Laser power, gain, pinhole and offset were set at the beginning of an imaging session and not adjusted thereafter. Stacks were deconvolved (Huygens Professional 17.10, Scientific Volume Imaging B.V.) to minimize the effect of the point spread function. We imported deconvolved images into Reconstruct (SynapseWeb, RRID:SCR_002716; Fiala, 2005). From there, we exhaustively sampled all tracer-labeled terminations within each stack, circling them to obtain their major diameter, and determining the proportion apposed to aspiny or sparsely spiny dendritic shafts immunolabeled for PV, CB, or CR. Appositions were defined as a close contact between the tracer and a CBP-labeled structure, often visible by a point of colocalization. The proportion apposed to presumed excitatory structures was computed as the remainder after subtraction of proportion apposed to PV, CB, and CR.
Synaptic analysis
After processing sections for EM synaptic analysis, we dissected small cubes from BL. We sampled from BLmc (basolateral magnocellular division) and BLi (basolateral intermediate division) for A25 and BLpc (basolateral parvocellular division) for hippocampus, where boutons were plentiful. These cubes were placed in a drop of fresh resin on premade LX112 resin coffin blocks (A25, case BQ, BW; hippocampus, case BS). These blocks were then cured for ≥48 h at 60°C. We then used an ultramicrotome (Ultracut UCT, Leica Microsystems) to cut ∼50-nm ultrathin sections and collected them sequentially on pioloform-coated copper slot grids to form series of 20–200 sections. For block-face imaging (hippocampus, case BQ), dissected cubes of tissue were affixed to aluminum pins using conductive epoxy glue (Chemtronics, catalog #CW2400). We used an ultramicrotome to expose the surface of the tissue in the resin blocks or pins. After exposing the tissue, pins for block-imaging were painted with conductive silver paint (Ted Pella, catalog #16035), which reduces charging artefacts in the scanning electron microscope (SEM).
We imaged specimens using one of two microscopes. The first is an 80 kV transmission electron microscope (TEM; 100CX, JEOL; A25, case BQ, BW; hippocampus, case BS). Boutons that had been labeled with DAB were located systematically and photographed for analysis using a digital camera (DigitalMicrograph, Gatan). EM stacks were aligned manually in Reconstruct.
For block-face imaging (hippocampus, case BQ), we mounted pins into the 3View 2XP System (Gatan) coupled to a 1.5 kV SEM (GeminiSEM 300, Zeiss). The surface of the pin was imaged using a backscatter detector, and then regions of interest (ROIs) were selected based on the presence of labeled axons. ROIs were generally 20- to 25- × 20- to 25-µm fields imaged at 6.5-nm resolution. A built-in microtome then cut a 50-nm section from the surface of the pin, and the ROIs were imaged again. In this way, long series of 100–300 sections were imaged in sequence at each ROI. For series obtained using block-face imaging, we used software for alignment (GMS3.0, Gatan). We imported EM stacks into Reconstruct (SynapseWeb, RRID:SCR_002716; Fiala, 2005).
Using Reconstruct, for both TEM-obtained or SEM-obtained stacks we exhaustively measured all DAB-labeled boutons for major diameter to quantify bouton size. We also noted several qualitative factors in the synaptic interaction (e.g., the presence or absence of mitochondria in a bouton). We selected one stack, obtained through block-face high throughput imaging, for 3D reconstruction (hippocampal case BQ). All labeled axons were traced, as were the synapses they formed, and the respective postsynaptic structures in entirety when possible. Reconstructions were rendered using 3Ds Max (Autodesk).
Data analysis and statistics
We used SPSS for data analysis and statistics (IBM, RRID:SCR_002865). Means were reported with SD (n = 2) or SEM (n > 2). We used one-way or two-way ANOVA with post hoc pair-wise Bonferroni tests to determine whether there were differences among stereological estimations of the density of terminations across nuclei, for bouton size analysis, and for analysis of appositional and synaptic targets. We used t tests for paired comparison of bouton size analysis, and some synaptic analysis data (e.g., presence of mitochondria in A25 vs hippocampal boutons in BL).
Brightness, contrast, and saturation adjustments for photographs were conducted using ImageJ (RRID:SCR_003070; Rasband, 1997–2014) or Adobe Photoshop (RRID:SCR_014199) but no retouching was performed. We prepared figures in Adobe Illustrator CC (RRID:SCR_010279).
Results
We used macro- to micro-scale quantitative mapping techniques to study pathways from the hippocampus and A25 to the amygdala (Fig. 1). After injection of anterograde neural tracers into hippocampus or A25, we used immunohistochemistry to label these pathways and study their postsynaptic interactions in amygdala using brightfield, confocal, and electron microscopy (EM). To delineate the borders of nuclei in the amygdala we used matched series of sections stained for Nissl (Fig. 2A), acetylcholinesterase (AChE; Fig. 2B), or unstained wet-mounted fixed tissue (Fig. 2C), in which white matter tracts appear dark, as described previously (Timbie and Barbas, 2015). We parcellated the amygdala according to previous studies (Barbas and De Olmos, 1990; De Olmos, 1990).
Experimental design. A, Schematic of medial view of rhesus monkey brain, depicting injection of neural tracers in subgenual cingulate A25 and the hippocampus. B, Schematic depicts the action of anterograde neural tracing. Neurons at the injection site incorporate the dye (here, green) and propagate it through their axons, producing labeled axon terminations in connected brain regions. C, Schematic of a mid-level section through the amygdala. D, Schematic depicts stereologic sampling of tracer-labeled boutons in the intercalated masses (IM). The optical fractionator method drops a grid over the structure, then drops a counting frame, or disector, within certain grid spaces applied in a systematic random fashion according to a preset sampling rate. E, At higher resolution, we study the postsynaptic targets of axonal terminations. F, Confocal photomicrograph of field of view in the basolateral nucleus (BL) of the amygdala depicts neural tracer labeled axons (green) and an inhibitory neuron marker (red). By analyzing rates of apposition, or points of close contact, for tracer-labeled elements and labeled dendritic elements, we can ascertain with a large sample size the targeting patterns of A25 and hippocampus in amygdala. G, Electron photomicrograph depicts a tracer-labeled axon termination forming a synapse (white arrow) on a dendritic shaft. Together, panels C–G depict our method of mapping system-level connectional patterns and using increasingly high-resolution methods to examine postsynaptic targets of pathways. A, anterior; Amy, amygdala; BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala (also known as accessory basal); CB, calbindin; Co, cortical nucleus of the amygdala; CR, calretinin; D, dorsal; Hipp, hippocampus; IM, intercalated masses of the amygdala; L, lateral; La, lateral nucleus of the amygdala; M, medial; Me, medial nucleus of the amygdala; P, posterior; PLBL, paralaminar basolateral nucleus of the amygdala; PV, parvalbumin; V, ventral.
Amygdala architecture. A, Photomicrograph of a coronal section through the mid-level of the amygdala (case AY), stained for Nissl. Nuclear boundaries delineated by black lines. Compass and scale bar apply to B, C. B, Photomicrograph of a coronal section through the mid-level of the amygdala (case BT), stained for acetylcholinesterase. Nuclear boundaries are delineated by white lines. C, Photomicrograph of a coronal section of wet-mounted fresh tissue matched to B (case BT), in which white matter appears dark. Nuclear boundaries are delineated by black lines. AHA, amygdalohippocampal transition area of the amygdala; BL, basolateral nucleus of the amygdala; BLmc, magnocellular division of BL; BLpc, parvocellular division of BL; BM, basomedial nucleus of the amygdala; CeL, lateral division of the central nucleus of the amygdala; CeM, medial division of the central nucleus of the amygdala; Co, cortical nucleus of the amygdala; D, dorsal; IM, intercalated masses of the amygdala; L, lateral; La, lateral nucleus of the amygdala; M, medial; Me, medial nucleus of the amygdala; P, posterior; PLBL, paralaminar basolateral nucleus of the amygdala; V, ventral.
Injection sites in hippocampus and A25
Figure 3 depicts the injection sites of neural tracers, which included A25 (Fig. 3A) and hippocampus (Fig. 3B) in six cases (n = 5 female; Table 1). In A25, the five injection sites ranged from anterior to posterior and medial to orbital sectors of A25. One injection (BW, midorbital A25; Fig. 3C) is reported here for the first time, while the rest have been used for previous unrelated studies (Fig. 3D–G; Joyce and Barbas, 2018; Joyce et al., 2022). The hippocampal subfields included in the injection sites varied by case (Fig. 3B,H–K), but all four injection sites included the CA1 subfield (Table 1). Injections in the hippocampus focused on the anterior hippocampus, as previous studies have demonstrated weak to sparse connectivity of the posterior hippocampus with the amygdala (Aggleton, 1986; Saunders and Rosene, 1988; Amaral and Insausti, 1992; Fudge et al., 2012). Moreover, the anterior hippocampus is the predominant origin of hippocampal projections to the anterior cingulate cortex, including subgenual A25, in rhesus monkeys (Barbas and Blatt, 1995; Wang et al., 2021). Together, these findings justify comparing the pathways from anterior hippocampus and A25 in the amygdala.
Injection sites in A25 and hippocampus. A, Medial (top) and orbital (bottom) surfaces of the rhesus monkey brain depict a normalized A25 space with injection sites. Dashed lines depict areal boundaries according to the map of Barbas and Pandya (1989). B, Photomicrographs of coronal sections through the hippocampus stained with Acetylcholinesterase depict locations of injection sites. Lines delineate hippocampal subregions. Compass applies to later hippocampus panels H–K. C, Brightfield photomicrograph of wet-mounted coronal section through A25 shows injection site (white arrowhead) of Cascade Blue in orbital A25. Dashed line depicts white matter (as in panels C–G); solid line depicts boundary with anterior olfactory nucleus; hatching depicts tissue rip. D, Fluorescent photomicrograph of coronal section through A25 shows injection site (white arrowhead) of Fluoroemerald in orbital A25. E, Darkfield photomicrograph of coronal section through A25 shows injection site (white arrowhead) of Lucifer yellow in medial A25. F, Fluorescent photomicrographs of two coronal sections through A25 show injection site (white arrowhead) of Fluororuby in medial A25. G, Brightfield photomicrograph of coronal section through A25 shows injection site (white arrowheads) of biotinylated dextran amine (brown) in posterior A25. The section is counterstained for Nissl (blue). H, Fluorescent photomicrograph of coronal section through hippocampus shows an injection site (white arrow) of Alexa488 in case BQ. The injection includes the dentate gyrus, CA1, prosubiculum, and subiculum (Table 1). I, Fluorescent photomicrograph of coronal section through hippocampus shows an injection site (white arrow) of Cascade Blue in CA1 of case BS. J, Fluorescent photomicrograph of coronal section through hippocampus shows an injection site (white arrow) of Fluoroemerald in case BT. The injection includes CA3, dentate gyrus, and CA1. K, Fluorescent photomicrograph of coronal section through hippocampus shows an injection site of Fluoroemerald in case BW. The injection site pictured includes CA3 and dentate gyrus. This injection site also included CA1 in more anterior levels (Table 1). Scale bars in panels C–K, 1 mm. A, anterior; All, allocortex; AON, anterior olfactory nucleus; BDA, biotinylated dextran amine; CA, Cornu Ammonis; CBL, Cascade Blue; cd, caudate; D, dorsal; DG, dentate gyrus; FE, Fluoroemerald; FR, Fluororuby; L, lateral; LOT, lateral olfactory tract; LY, Lucifer yellow; M, medial; MPAll, Medial PeriAllocortex; P, posterior; PreS, presubiculum; ProS, prosubiculum; OPAll, Orbital PeriAllocortex; OPro, Orbital Proisocortex; os, olfactory sulcus; st, striatum; Sub, subiculum; uCA, uncal Cornu Ammonis; V, ventral.
A25 and hippocampal pathways terminate in some common amygdalar sites, and some divergent sites
To map termination patterns in amygdala, we traced bouton-bearing axons at high resolution, providing qualitative maps of axonal innervation in representative sections throughout the amygdala, as depicted for the hippocampus in Figure 4 (three sections/case, two cases). The hippocampus preferentially innervated the ventral aspects of the BL (the parvocellular subdivision, BLpc), basomedial (BM; also known as accessory basal), and the Cortical nucleus (Co). We also found very strong axonal labeling in PLBL, a granular nucleus at the most ventral aspect of the amygdala, in accord with previous findings (deCampo and Fudge, 2012; Fudge et al., 2012).
Hippocampal axon termination patterns in amygdala. A, Injection site in hippocampus (BQ-Alexa488) used for axon termination maps in panel B and photography in panels C–G. B, Individual axons (black) labeled from injection site in panel A traced at 1000× on three coronal sections through the amygdala. A qualitative assessment of density was used to apply modular axonal innervation patterns, adjusted for each field of view, for densely labeled sites in the ventral BM, BL, and PLBL (middle and right sections). Compass also applies to H. C, Low-power photomicrograph of the amygdala with tracer visualized by DAB (brown). D-E, Insets depict higher magnification photomicrographs of the ventral amygdala. F, Inset from E depicts a collapsed z-stack (0.5-µm step size, 1000×) of photomicrographs taken in the ventral amygdala. Double-headed arrows point out labeled axon terminations (brown) also shown in single focal plane (G) below. H, Injection site in hippocampus (BT-FE) used for axon termination maps in panel I. I, Individual axons (black) labeled from injection site in panel H traced at 1000× on three coronal sections through the amygdala. AHA, amygdalohippocampal transition area of the amygdala; BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala; CA, Cornu Ammonis; CeL, lateral division of the central nucleus of the amygdala; CeM, medial division of the central nucleus of the amygdala; Co, cortical nucleus of the amygdala; D, dorsal; DAB, diaminobenzidine; IM, intercalated masses of the amygdala; L, lateral; La, lateral nucleus of the amygdala; M, medial; Me, medial nucleus of the amygdala; P, posterior; PLBL, paralaminar basolateral nucleus of the amygdala; ProS, prosubiculum; V, ventral.
Terminations from A25 interacted with diverse amygdalar nuclei and revealed a connectional gradient reflecting the origin of pathways in medial (Fig. 5) or orbital (Fig. 6) sites within A25. All A25 sites targeted dorsal aspects of the amygdala (three sections traced/case, three cases; Figs. 5A,F, 6C,E), but with clear variation in targeting patterns emerging across A25 origin sites. To show the full breadth of complexity, we overlaid individually traced axons with maps of stereologically sampled axon terminations, or boutons, from the same coronal slices (five sections stereologically sampled per case; dots represent individual boutons in Figs. 5A,F, 6E). These three cases confirmed that the orbital sector of A25 had a somewhat different relationship with the amygdala than the medial sites of A25. To explore this further, we performed supplemental axon tracing in two additional cases (one section traced/case; Figs. 5C, 6C). Together, Figures 5 and 6 depict qualitative maps of individual traced axons and stereologically sampled boutons originating from these five sites within A25.
Medial A25 axon innervation patterns in amygdala. A, Coronal sections through amygdala (case BR-BDA) with individual axons (black lines) from injection site (in panel B) traced at 1000×, and stereologically sampled axon terminations overlaid (small dark gray dots in grid formation). Compass in A5 applies to all panels with amygdala tracings. White circle in A3 denotes location of photograph in D. B, Medial surface of the rhesus monkey brain, shows medial injection sites in A25 used to plot panels A, C, F. C, Coronal section in mid-level of amygdala (case BQ-FR) with individual axons (black lines) from injection site (in panel B) traced at 1000×. D, Collapsed brightfield photomicrograph z-stack (step size 0.5 µm) of ventral BM (case BR-BDA) shows tracer-labeled axons (brown; black arrows) and counterstaining for Nissl (blue). Photomicrograph location is shown in A3 (white circle). E, Inset from D shows single focal plane with labeled axons (brown; black arrows). F, Coronal sections through amygdala (case BS-LY) with individual axons (black lines) from injection site (in panel B) traced at 1000×, and stereologically sampled axon terminations overlaid (small dark gray dots in grid formation). AHA, amygdalohippocampal transition area of the amygdala; BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala; CeL, lateral division of the central nucleus of the amygdala; CeM, medial division of the central nucleus of the amygdala; Co, cortical nucleus of the amygdala; D, dorsal; IM, intercalated masses of the amygdala; L, lateral; La, lateral nucleus of the amygdala; M, medial; Me, medial nucleus of the amygdala; P, posterior; PLBL, paralaminar basolateral nucleus of the amygdala; V, ventral.
Orbital A25 axon innervation patterns in the amygdala. A, Collapsed brightfield photomicrograph z-stack (step size 0.5 µm) of BM (case BW-CBL) shows tracer-labeled axons (brown; black arrows). Location of photomicrograph is denoted in C with a white circle. B, Inset from A shows single focal plane with labeled axons (brown; black arrows). C, Coronal section in mid-level of the amygdala (case BW-CBL) with individual axons (black lines) from injection site (in panel D) traced at 1000×. D, Orbital surface of the rhesus monkey brain, shows orbital injection sites in A25 used to plot panels C, E). E, Coronal sections through the amygdala (case BU-FE) with individual axons (black lines) from injection site (in panel D) traced at 1000×, and stereologically sampled axon terminations overlaid (small dark gray dots in grid). Compass in E5 also applies to all other amygdala tracings. AHA, amygdalohippocampal transition area of the amygdala; BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala; CeL, lateral division of the central nucleus of the amygdala; CeM, medial division of the central nucleus of the amygdala; Co, cortical nucleus of the amygdala; D, dorsal; IM, intercalated masses of the amygdala; L, lateral; La, lateral nucleus of the amygdala; M, medial; Me, medial nucleus of the amygdala; P, posterior; PLBL, paralaminar basolateral nucleus of the amygdala; V, ventral.
Terminations from medial sectors of A25 in amygdala are depicted in Figure 5. The medial sectors of A25 heavily innervated BM (Fig. 5A2–A4,C,F4,F5), the medial nucleus (Me; Figs. 5A1–A4,F4,F5), and the central nucleus (Ce; Fig. 5A3–A5,F4). Medial A25 innervated all subcompartments of BL, including the dorsal sector populated by large neurons called the magnocellular division (BLmc; Fig. 5A4,F4), BLi (Fig. 5A4,F3,F4), and more posterior aspects of the ventral BLpc (Fig. 5A3,A4,F4). Pathways from medial A25 also terminated heavily in ventral La (Fig. 5A3,A4,F3,F4). The posterior medial sector of A25 additionally heavily innervated the amygdalohippocampal transition area (AHA) and the ventral BLpc (AHA; Fig. 5A3–A5).
Figure 6 shows terminations from orbital sectors of A25 in amygdala. The orbital sector of A25 was distinguished by heavy innervation of the intercalated masses (IM) in a striking pattern not seen for pathways from medial sites of A25 or from hippocampus (Fig. 6C,E2,E3). Orbital A25 abuts the posterior orbitofrontal cortex (pOFC), which also preferentially innervates the IM in a similarly striking pattern (Ghashghaei and Barbas, 2002; Zikopoulos et al., 2017). Orbital A25 is sometimes classified as a distinct cytoarchitectonic area (area 14c in another map; Carmichael and Price, 1994), however we find its cytoarchitectonic and chemoarchitectonic features to be more akin to A25 than area 14, and thus we call it orbital A25 (for discussion, see Joyce and Barbas, 2018). Orbital A25 also innervated the Me, Ce, and Co nuclei and dorsal sectors of BL (Fig. 6C,E2–E4), reminiscent of pathways from the nearby medial sectors of A25. The most posterior part of orbital A25 displayed this pattern more prominently (Fig. 6C). These results suggest that A25 has a complex and diverse array of interactions in the amygdala.
Both hippocampal and A25 pathways innervated the ventral aspect of BL, where neurons are smaller, called the parvocellular division (BLpc; Figs. 4B,I, 5A3,A4,F4). However, hippocampal pathways tended to terminate in more anterior and medial BLpc (Fig. 4), while A25 axons terminated in more posterior and lateral BLpc (Fig. 5C,F3,F4). Our maps suggest that terminations from A25 and hippocampus overlap in BL and especially in the middle levels of the amygdala in the rostro-caudal dimension.
Figure 7 depicts the results of stereological quantitative analysis (three cases per pathway, five sections per case). Across amygdalar nuclei, BL received the definitive bulk of terminations from both A25 and hippocampus (45 ± 5% boutons from A25; 57 ± 2% of boutons from hippocampus; Fig. 7A1). This was statistically significant within pathways (A25: one-way ANOVA, F(9,20) = 18.1, p < 0.001, post hoc Bonferroni, BL vs all other nuclei p < 0.001; hipp: one-way ANOVA, F(8,18) = 135.5, p < 0.001, post hoc Bonferroni, BL vs all other nuclei p < 0.001). Some differences across pathways were also significant (one-way ANOVA, F(17,36) = 33.8, p < 0.001, post hoc least significant difference; Fig. 7A, asterisk).
Quantitative mapping of A25 and hippocampal axon terminations in nuclei of the amygdala. A, Mean percent of axon terminations from A25 (green) or hippocampus (blue) in amygdalar BL; A1 shows values for BL, which had the highest proportion of boutons in the amygdala, and the two intrinsic nuclei of the amygdala, IM (innervated by orbital A25) and PLBL (innervated by the hippocampus); A2 shows values for all other nuclei with label in the amygdala. Individual cases are denoted by color-coded shapes. B, Mean relative volumetric density of A25 (green) and hippocampal (blue) boutons found in amygdalar nuclei. For each case, the volumetric density of individual amygdalar nuclei was divided by the volumetric density of each case's the most densely innervated nucleus to produce a normalized index. *p < 0.05, significance shown only for between pathway comparisons. AHA, amygdalohippocampal transition area of the amygdala; BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala; CA, Cornu Ammonis; CeL, lateral division of the central nucleus of the amygdala; CeM, medial division of the central nucleus of the amygdala; Co, cortical nucleus of the amygdala; DG, dentate gyrus; Hipp, hippocampus; IM, intercalated masses of the amygdala; La, lateral nucleus of the amygdala; M, medial; Me, medial nucleus of the amygdala; PLBL, paralaminar basolateral nucleus of the amygdala; ProS, prosubiculum; Sub, subiculum.
We then used stereological results to compute the volumetric density (boutons/µm3) by nucleus. For each case, we normalized the density of each nucleus to the most densely innervated nucleus to produce a relative density index. Across the hippocampal cases, PLBL exhibited the highest density index in innervation from hippocampal boutons (Fig. 7B1). This measure produced varied results across medial and orbital A25 subsectors, reminiscent of the connectional gradient (Fig. 7B1). Some differences across pathways were significant (one-way ANOVA, F(17,36) = 6.1, p < 0.001, post hoc least significant difference; Fig. 7B, asterisk). Bouton density analysis thus confirmed the diverse impact of A25 on amygdala nuclei, while underscoring a notable pairing between hippocampus and PLBL.
A25 terminations were larger than hippocampal terminations in the basolateral complex
Because the bulk of terminations from A25 and hippocampus fell in BL (Fig. 7A1), we then used higher resolution methods to compare the size of terminations from the two pathways in BL (Fig. 8). We performed this analysis because bouton size is correlated with synaptic efficacy (Stevens, 2004; Rollenhagen and Lübke, 2006). Using immunohistochemistry for fluorescence and EM, we captured confocal z-stacks and serial EM images of tracer-labeled axon terminations and measured their major diameter. Confocal analysis featuring a large sample size revealed that A25 boutons were significantly larger than hippocampal boutons in BL (A25: 0.84 ± 0.02 µm; hipp: 0.70 ± 0.05 µm; t(5) = 3.8, p = 0.007; Fig. 8C,D). We also saw this trend using EM, although it was not statistically significant likely because of the smaller sample sizes that are feasible for this high-resolution analysis.
Bouton diameter of A25 and hippocampal axon terminations in the amygdala. A, Fluorescent photomicrograph of axon terminations (green) in amygdala comingled with local PV inhibitory neuronal processes (red). B, Inset from A. Axon terminations are circled (white outlines) to obtain the major (largest) diameter. C, Mean distribution of major diameters from A25 and hippocampal boutons in BL of amygdala obtained by confocal microscopy with deconvolution. D, Mean bouton major diameter of A25 and hippocampus in BL obtained by confocal microscopy. BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala; Hipp, hippocampus; La, lateral nucleus of the amygdala; PLBL, paralaminar basolateral nucleus of the amygdala; PV, parvalbumin.
We extended the bouton size analysis using confocal microscopy to other nuclei of the BLcx that were innervated by A25 and hippocampus. In the other BLcx nuclei examined, bouton diameter of terminations from A25 and hippocampus were consistent with findings in BL (A25 to BM: 0.81 ± 0.09 µm, n = 3 cases, 13 014 boutons; A25 to La: 0.78 ± 0.08 µm, n = 3 cases, 8295 boutons; hippocampus to PLBL: 0.66 ± 0.01 µm, n = 2 cases, 6361 boutons).
Pathway interactions with inhibitory neurons in the amygdala: A25 and hippocampal terminations innervated CR neurons, but diverged in the order of targeting PV and CB neurons in BLcx
We then studied the relationship of hippocampal and A25 pathways with inhibitory neurons in BL, where the two pathways terminate in common. Inhibitory neurons make up a small proportion of neurons in the BLcx (∼10–20%), but they can have outsized effects on the functional architecture (McDonald and Augustine, 1993; Ehrlich et al., 2009; Spampanato et al., 2011; Lee et al., 2013; Bazelot et al., 2015; Krabbe et al., 2018). As in the primate cortex, PV, CB, and CR are reliable markers for grouping amygdalar inhibitory neurons into mostly distinct neurochemical and functional classes in the BLcx. We double immunolabeled for tracer and PV, CB, or CR (Fig. 9A–F) to study appositions between tracer-labeled terminations and putative inhibitory dendrites in the BLcx, to determine the interaction of each pathway with specific neurochemical classes of inhibitory neurons. By analyzing the rates of apposition (Fig. 9D–F, white arrowheads) between tracer labeled axons and PV, CB, or CR dendritic shafts, we captured a portrait of each pathway's putative synaptic targeting patterns with each neurochemical class of inhibitory neurons.
Apposition analysis of A25 and hippocampal terminations in amygdala. A-C, Collapsed z-stacks (maximum projection) of fluorescent photomicrographs show A25 terminations in BL comingling with PV (A), CB (B), or CR (C) neuronal processes. Scale bar applies to A–C. D-F, Higher magnification collapsed z-stacks (maximum projection) of fluorescent photomicrographs show hippocampal terminations in BL comingling with PV (D), CB (E), or CR (F) neuronal processes. Insets show rotated views of appositions (white arrowheads), or points of close contact, between axon terminations and neuronal processes. Scale bar applies to D–F. G, Color-coding shows A25 (green) and hippocampal (blue) pathways to BL in amygdala for panels H–J. H, Mean proportion of axon terminations that were apposed to inhibitory dendritic processes (black, % presumed inhibitory, summed over PV, CB, and CR), or axons that interacted with an unlabeled postsynaptic structure (gray; % unlabeled, likely largely excitatory, computed by subtraction) in amygdalar nuclei using confocal analysis. I, Breakdown of putative inhibitory appositions per pathway by PV, CB, or CR (shown in H) for BL using confocal analysis. J, Mean percent of synapses formed by A25 and hippocampal boutons in BL that interacted with sparsely spiny or aspiny dendrites positive for PV, CB, or CR (likely inhibitory) using EM synaptic analysis. *p < 0.05. BL, basolateral nucleus of the amygdala; BM, basomedial nucleus of the amygdala; CB, calbindin; CR, calretinin; Hipp, hippocampus; La, lateral nucleus of the amygdala; n.s., not significant; PLBL, paralaminar basolateral nucleus of the amygdala; PV, parvalbumin.
In BL, A25 terminations were apposed to putative inhibitory dendrites at a higher rate than hippocampal terminations (Fig. 9H). This trend was driven by A25 axon terminations being apposed to significantly higher levels of CR and PV neurons than the hippocampus in BL (two-way ANOVA, F(14,24) = 9.7, p = 0.001; post hoc Bonferroni PV, A25 vs hipp: p = 0.003; CR, A25 vs hipp: p = 0.037; Fig. 9I). Both A25 and the hippocampus preferentially interacted with CR dendritic shafts, but then diverged in their targeting patterns with PV and CB neurons. After CR, A25 terminations preferentially apposed PV dendrites, and finally CB, while hippocampal terminations apposed CB dendrites and least PV sites (Fig. 9I). Appositional patterns were consistent in other BLcx nuclei as well. A25 terminations tended to appose more inhibitory dendritic shafts in BL than in BM and La (21 ± 1% in BL, 16 ± 2% in BM and 15 ± 0.3% La, respectively), but were otherwise consistent in order of inhibitory neurons targeted (A25 in BM: apposed to CR 10.6 ± 1%, to PV 4 ± 1%, to CB 2 ± 0.5%; A25 in La: apposed to CR 8 ± 1%, to PV 5 ± 1%, to CB 3 + 0.2%). Appositions by hippocampal terminations were consistent across BL and PLBL (Hippocampal in PLBL: apposed to CR 6 ± 2.5%, to CB 4.5 ± 1%, to PV 1 ± 0.4%).
Using serial EM, we performed a higher resolution analysis of the postsynaptic targets of long-range inputs from A25 and hippocampus forming synapses in BL, and the results are shown in Figure 9J. Corroborating the patterns seen from confocal appositional analysis, we saw the same trend in EM. Thus, among inhibitory neurons, A25 innervated CR dendritic shafts, followed by PV, and then CB. Hippocampal terminations also followed the same order seen with the confocal analysis, revealing the same trend, although not statistically significant.
Most terminations in BL formed synapses with spines on densely spiny dendrites, which are putatively excitatory (Fig. 10A–C,F–H), like the large proportion of appositions with unlabeled targets, which are presumed to be excitatory (Fig. 9H). A smaller proportion of boutons formed synapses with dendritic shafts that appeared aspiny (“smooth”) or sparsely spiny in serial EM (Fig. 10D,E,I,J), a characteristic of inhibitory neurons in BLcx and cortex. The majority of A25 and hippocampal boutons formed synapses on single spines of densely spiny, presumed excitatory, dendrites and the remaining boutons formed synapses on sparsely spiny dendritic shafts, or were multisynaptic boutons (Fig. 11A). Figure 11B–D shows an example of reconstruction from high throughput serial EM obtained by block-face imaging, featuring a hippocampal axon termination forming a synapse on a CB+ smooth dendrite (red), and another axon forming a synapse with the spine of a pyramidal-like putative excitatory dendrite (green).
Photomicrographs of synapses formed by hippocampus and A25 in the basolateral nucleus of the amygdala. A, B, A hippocampal axon termination (labeled with DAB, dark precipitate) forms a round, asymmetric synapse (excitatory, arrow) with a spine (sp, green) in BL of amygdala. Scale bar applies to all panels. C, A multisynaptic hippocampal axon termination forms asymmetric synapses (round, arrow; perforated, double-headed arrows) with two spines (green). D, A hippocampal axon termination forms an asymmetric synapse (arrow) on a dendritic shaft (outlined in orange) with gold labeling for CB (appears as black cluster). E, A hippocampal axon termination forms an asymmetrical synapse (arrow) on a dendritic shaft (orange outline) with TMB for CR (appears as black patchy precipitate). F, An A25 axon termination forms an asymmetric synapse (arrow) with a spine (green) in BL of amygdala. G, Two distinct A25 axonal terminations form asymmetric synapses (arrows) with two distinct spines (green). H, An A25 axon termination forms an asymmetric synapse (arrow) with a spine (green) in BL of amygdala. I, An A25 axon termination forms an asymmetric synapse (arrow) with a dendritic shaft (outlined in orange) labeled with CB (gold, deep black clusters). J, An A25 axon termination forms an asymmetric synapse (arrow) with a dendritic shaft (orange outline) labeled with CR (gold, black cluster). BL, basolateral nucleus of amygdala; CB, calbindin; CR, calretinin; DAB, diaminobenzidine; mito, mitochondria; sp, spine; TMB, tetramethylbenzidine.
Synaptic analysis of hippocampal and A25 terminations in the BL nucleus of the amygdala. A, % of boutons that formed synapses with single spines, dendritic shafts, or combinations thereof, in BL analyzed from each case (A25, left; hippocampus, right). B, Reconstruction created from high-throughput serial EM depicting hippocampal axons (blue) in BL forming synapses with a spine emanating from a pyramidal-like dendrite (green), and a smooth CB+ putatively inhibitory dendrite (red). Analyzed region of interest is 26 × 26 µm with a z-depth of 7.25 µm. C1, Rotated image from B. C2, C3, Inset from B, rotated to depict a hippocampal axon (blue) forming a synapse (black arrow, yellow) with a smooth CB+ dendrite (red). D1, Slightly rotated image from B. D2, D3, Inset from B, rotated to depict a hippocampal axon (blue) forming a synapse (black arrow, yellow) with a spine emanating from a spiny putative pyramidal dendrite. Scale cube, 1 µm3. BL, basolateral nucleus of the amygdala; CA, Cornu Ammonis; Hipp, hippocampus; n.s., not significant; Sub, subiculum.
Other relevant synaptic characteristics showed that A25 had a higher tendency to form multisynaptic terminations (A25: 9 ± 1.4%, hipp: 6 ± 3%), to contain mitochondria (A25: 66 ± 2%, hipp: 58 ± 22%), to interact with a spine containing a spine apparatus (A25: 30 ± 2%, hipp: 23 ± 16%), although none of these distinctions were statistically significant. The presence of mitochondria may signal a more active synapse (for review, see Devine and Kittler, 2018). Spines containing smooth endoplasmic reticulum, or a spine apparatus, are associated with plasticity and calcium dynamics (for review, see Segal et al., 2010). Synapse shape can also have an impact on synaptic function. Perforated synapses are associated with a larger postsynaptic density and correlate with AMPA receptor density (Geinisman, 1993). A25 and hippocampal boutons formed perforated synapses at comparable rates (A25: 19 ± 3%, hipp: 21 ± 13%).
Discussion
The amygdala received common innervation in the BL nucleus from hippocampus and A25, as well as diametrically divergent projections to two distinct nuclei that affect the internal processing of the amygdala (Fig. 12A). The hippocampus and A25 preferentially innervated CR inhibitory neurons but diverged in targeting CB and PV inhibitory neurons (Fig. 12B). This circuit specificity demonstrates that the hippocampus and A25 are positioned to exert collaborative as well as specialized roles in the amygdala, in processes of affective significance for flexible behavior which may be disrupted in psychiatric diseases.
Summary of terminations in amygdala. A, Diagram of a coronal section through a mid-level of the amygdala. Hatching depicts termination zones for A25 (medial, blue stripes; orbital, red dots) and hippocampus (green grid). Hatching color saturation indicates a coarse representation of termination density. B, Simplified schematic of predominant postsynaptic targeting patterns of A25 and hippocampal terminations in BL. Shades of green indicate excitatory neuronal processes; shades of red and orange indicate inhibitory neuronal processes. A25 and hippocampal inputs are coarsely weighted. Both A25 and hippocampus heavily innervated spines of densely spiny pyramidal-like dendrites, followed by CR putative inhibitory smooth dendrites. Hippocampal afferents then innervated CB putative inhibitory smooth dendrites, while A25 afferents innervated PV putative inhibitory smooth dendrites. BL, basolateral nucleus of the amygdala; BLmc, magnocellular division of BL; BLpc, parvocellular division of BL; BM, basomedial nucleus of the amygdala; Ce, central nucleus of the amygdala; Co, cortical nucleus of the amygdala; CB, calbindin; CR, calretinin; Hipp, hippocampus; IM, intercalated masses of the amygdala; La, lateral nucleus of the amygdala; Me, medial nucleus of the amygdala; PLBL, paralaminar basolateral nucleus of the amygdala; PV, parvalbumin.
Common and divergent hippocampal and A25innervation of the amygdala
The anterior hippocampus and A25 commonly innervated the BL and BM nuclei of the amygdala. The hippocampal pathway projected heavily to the ventromedial parts of BL and BM nuclei, in a pattern that overlaps with pathway targets of medial A25. A25 had a broader innervation pattern in BL and BM, and also had a unique projection to the ventral part of the La nucleus. The BLcx receives topographic innervation from cortical sensory streams and is highly interconnected (for review, see LeDoux, 2007; McDonald, 2020). Neurons in BLcx are thought to integrate sensory stimuli with positive or negative valence (Zhang et al., 2013; Janak and Tye, 2015), suggesting a wider role beyond fear-related processes that are so salient in psychiatric disturbances. La is most known for studies of auditory fear conditioning in rats (Romanski et al., 1993) and also receives inputs from other unimodal sensory streams in macaques (McDonald, 2020). The BM in macaques receives mostly polymodal input (McDonald, 2020).
The hippocampus and A25 had unique interactions with other amygdalar nuclei. The Me nucleus, which is associated with olfactory functions (De Olmos, 1990), and the amygdalar output-associated central nucleus, were densely innervated by A25 but only sparingly by the hippocampus. The hippocampus innervated the ventral Co nucleus, which is associated with endocrine functions (De Olmos, 1990), more densely than A25, although A25 terminations spanned the dorsoventral extent of the nucleus. These findings suggest that A25 has broad influence on diverse aspects of processes of the amygdala and its output (see also Ghashghaei et al., 2007), while hippocampal influence appears to be more circumscribed.
There was a major divergence in the innervation patterns of A25 and the hippocampus of the intrinsic amygdalar nuclei. Orbital A25 was distinguished by strong innervation of the entirely inhibitory striatal-like IM, in a pattern similar to adjacent pOFC areas (Ghashghaei and Barbas, 2002; Zikopoulos et al., 2017). Situated in narrow strips between the basal and central nuclei, the primate IM form a complex multisynaptic network of inhibition with chemoarchitecture that is distinct from the basal nuclei (Zikopoulos et al., 2016). In rats, prefrontal projections to IM are associated with fear inhibition and extinction (Likhtik et al., 2008; Hagihara et al., 2021; Seewald et al., 2021). The connectional gradient seen from medial to orbital A25 may play a part in the markedly different physiological responses reported during aversive and appetitive behavioral paradigms in macaques (Monosov and Hikosaka, 2012).
The hippocampus had a different specialized entry to the amygdala via the PLBL, an enigmatic nucleus in the ventral aspect of the amygdala, which has expanded in primates (deCampo and Fudge, 2012; Fudge et al., 2012). The PLBL has limited extrinsic inputs, suggesting it could be largely dedicated to hippocampal afferents (deCampo and Fudge, 2012; Fudge et al., 2012). The PLBL receives intrinsic amygdalar connections from La and projects to both Ce and BL (Amaral and Insausti, 1992; Pitkänen and Amaral, 1998; Fudge and Tucker, 2009), suggesting a unique and separate route for the hippocampal pathway to influence CeM and downstream autonomic structures.
Hippocampal and A25 pathways innervate first CR inhibitory neurons but diverge in secondary preference for CB and PV inhibitory targets in amygdala
Most terminations from hippocampus and A25 innervated putatively excitatory postsynaptic targets (80–85%), while a significant proportion of synapses (∼15–20%) were on presumed inhibitory neurons in BL. Among inhibitory targets, both afferent pathways displayed preference for CR neurons. As in the upper layers of cortex, amygdalar CR neurons inhibit other inhibitory neurons and thus are thought to have a disinhibitory role (Sorvari et al., 1998; Muller et al., 2003). BLcx nuclei are known for their low spontaneous firing rates likely because of tightly regulated inhibitory networks (for review, see Quirk and Gehlert, 2003; Perumal and Sah, 2021). Disinhibition via CR neurons may be a permissive gate for coincident detection of pertinent signals based on mood state or context (Wolff et al., 2014; Letzkus et al., 2015; Krabbe et al., 2018, 2019).
Beyond their preferential targeting of CR neurons, innervation patterns of other inhibitory targets diverged. A25 innervated a higher proportion of the powerful PV inhibitory neurons than the hippocampus. Innervation of PV neurons, which target perisomatic sites of nearby pyramidal neurons, may allow A25 to rapidly dampen excitatory activity (for review, see Hajos, 2021). Parvalbumin inhibitory neurons in BL are an important component of fear inhibition and extinction processes in amygdala (Chhatwal et al., 2005; Heldt et al., 2012; Trouche et al., 2013; Saha et al., 2017; for review, see Saha et al., 2020; Hajos, 2021). Synapses with PV neurons in macaques suggest a role for A25 to rapidly adjust gain in sensory-valence association assemblies, in support of behavioral flexibility in nonpathologic mood states.
We found the opposite trend in the innervation of CB inhibitory neurons: the hippocampal pathway innervated a higher proportion of CB putative inhibitory postsynaptic sites in the amygdala than A25. CB neurons, which are often somatostatin-positive, innervate mid to distal dendrites of pyramidal neurons, forming inhibitory synapses very proximal to excitatory inputs in the dendritic tree (Muller et al., 2007; Chiu et al., 2013; McDonald, 2020; Hajos, 2021). CB neurons are thus poised to offset discrete excitatory synapses in the dendritic summation process. Recent studies have shown that reduced somatostatin neuron activity in the amygdala can facilitate associative memory formation and retrieval (Wolff et al., 2014; Krabbe et al., 2019; Hajos, 2021), which highlights the possibility of a delicate balance struck by hippocampal afferents for flexible, contextually based behavior.
Functional implications
Our findings suggest collaborative as well as distinct influences by hippocampal and A25 pathways in the primate amygdala. Hippocampal engagement of CB neurons may help maintain specificity in associative neuronal assemblies and guard against overgeneralization, in line with findings that the hippocampus signals to prefrontal cortex which associations are correct during learning (Brincat and Miller, 2015). Further, the hippocampus densely innervated the PLBL nucleus, which has been associated with immature-appearing neurons, suggesting high plasticity (deCampo and Fudge, 2012; Fudge et al., 2012). The hippocampal-PLBL pairing may play a specialized and crucial role for flexible affective-related learning processes in primates throughout life.
Overall, A25 appears to have a broader reach, a tendency toward stronger synaptic interactions in the amygdala, and denser direct influence on the CeM nucleus, suggesting extensive involvement in amygdalar influence on autonomic structures. Posterior A25, which featured a hyper-dense termination plexus in the basal complex of the amygdala (e.g., Fig. 5A3,A4), is overactive in depression (Hamani et al., 2011). We speculate that dense terminations from this posterior A25 site on excitatory neurons and disinhibitory CR neurons in BL may lead to excessive activation of the amygdala and its output streams in conditions of negative affect and mood disorders. Posterior medial A25 is relatively unique among prefrontal areas in its strength in sending to, and receiving pathways from the amygdala (Ghashghaei et al., 2007). These strong reciprocal connections (Kim et al., 2018; Sharma et al., 2020; Calderazzo et al., 2021) may induce pathologic feedforward excitatory loops that are difficult to disengage in depression. In turn, high activation of the amygdala can monopolize the intrinsic circuitry of hippocampus through its strong and specialized connections and lead to overgeneralization, as suggested in a recent study (Wang and Barbas, 2018).
Finally, the robust projections from orbital A25 to the inhibitory IM could have a conditional regulatory role on CeM, based on previous findings that most inhibitory neurons in IM express the striatal-related phosphoprotein DARPP-32 (Zikopoulos et al., 2016). By analogy with its function in striatum, high levels of dopamine render DARPP-32 neurons hyperpolarized (Greengard, 2001; Svenningsson et al., 2004), and may ultimately blunt the input from posterior orbitofrontal and orbital A25 to IM. This evidence suggests that in states of high stress, when dopamine levels are high, prefrontal mobilization of IM may be ineffective to regulate autonomic activity.
In summary, our findings suggest concerted influence by A25 and the hippocampus. These pathways may allow linkage of stimuli to rewarding outcomes in normal behavior for learning, and in support of complex social and cognitive processes mediated by neurons in the amygdala (Saez et al., 2015; Wellman et al., 2016). Overactivity in A25, which is the hallmark of depression (Mayberg et al., 2005), may also heighten autonomic arousal through the pathways to amygdala shown here, and through descending projections to autonomic structures (An et al., 1998; Freedman et al., 2000; Chiba et al., 2001; Alexander et al., 2019), and overwhelm the hippocampal input to amygdala that can provide signals pertaining to appropriate context in behavior.
Footnotes
This work was supported by National Institutes of Health Grants R01MH057414 and R01MH117785 (to H.B.). We thank Jess Holz, MFA for electron microscopy and technical assistance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Helen Barbas at barbas{at}bu.edu
